{
  "affected": [
    {
      "ecosystem_specific": {
        "fix": "bump",
        "range_state": "fixed",
        "resource": "langchain-core",
        "resource_purl": "pkg:pypi/langchain-core@1.4.9",
        "upstream_fixed_in": "1.2.5"
      },
      "package": {
        "ecosystem": "Homebrew",
        "name": "langgraph-cli",
        "purl": "pkg:brew/langgraph-cli"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "0.4.31"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "database_specific": {
    "confidence": "high",
    "source": "matched",
    "strategy": "registry",
    "upstream_evidence": [
      {
        "ecosystem": "PyPI",
        "key": "pkg:pypi/langchain-core@1.4.9",
        "name": "langchain-core",
        "resource": "langchain-core",
        "strategy": "registry",
        "subject_version": "1.4.9"
      }
    ]
  },
  "details": "## Summary\n\nA serialization injection vulnerability exists in LangChain\u0027s `dumps()` and `dumpd()` functions. The functions do not escape dictionaries with `\u0027lc\u0027` keys when serializing free-form dictionaries. The `\u0027lc\u0027` key is used internally by LangChain to mark serialized objects. When user-controlled data contains this key structure, it is treated as a legitimate LangChain object during deserialization rather than plain user data.\n\n### Attack surface\n\nThe core vulnerability was in `dumps()` and `dumpd()`: these functions failed to escape user-controlled dictionaries containing `\u0027lc\u0027` keys. When this unescaped data was later deserialized via `load()` or `loads()`, the injected structures were treated as legitimate LangChain objects rather than plain user data.\n\nThis escaping bug enabled several attack vectors:\n\n1. **Injection via user data**: Malicious LangChain object structures could be injected through user-controlled fields like `metadata`, `additional_kwargs`, or `response_metadata`\n2. **Class instantiation within trusted namespaces**: Injected manifests could instantiate any `Serializable` subclass, but only within the pre-approved trusted namespaces (`langchain_core`, `langchain`, `langchain_community`). This includes classes with side effects in `__init__` (network calls, file operations, etc.). Note that namespace validation was already enforced before this patch, so arbitrary classes outside these trusted namespaces could not be instantiated.\n\n### Security hardening\n\nThis patch fixes the escaping bug in `dumps()` and `dumpd()` and introduces new restrictive defaults in `load()` and `loads()`: allowlist enforcement via `allowed_objects=\"core\"` (restricted to [serialization mappings](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/load/mapping.py)), `secrets_from_env` changed from `True` to `False`, and default Jinja2 template blocking via `init_validator`. These are breaking changes for some use cases.\n\n## Who is affected?\n\nApplications are vulnerable if they:\n\n1. **Use `astream_events(version=\"v1\")`** \u2014 The v1 implementation internally uses vulnerable serialization. Note: `astream_events(version=\"v2\")` is not vulnerable.\n2. **Use `Runnable.astream_log()`** \u2014 This method internally uses vulnerable serialization for streaming outputs.\n3. **Call `dumps()` or `dumpd()` on untrusted data, then deserialize with `load()` or `loads()`** \u2014 Trusting your own serialization output makes you vulnerable if user-controlled data (e.g., from LLM responses, metadata fields, or user inputs) contains `\u0027lc\u0027` key structures.\n4. **Deserialize untrusted data with `load()` or `loads()`** \u2014 Directly deserializing untrusted data that may contain injected `\u0027lc\u0027` structures.\n5. **Use `RunnableWithMessageHistory`** \u2014 Internal serialization in message history handling.\n6. **Use `InMemoryVectorStore.load()`** to deserialize untrusted documents.\n7. Load untrusted generations from cache using **`langchain-community` caches**.\n8. Load untrusted manifests from the LangChain Hub via **`hub.pull`**.\n9. Use **`StringRunEvaluatorChain`** on untrusted runs.\n10. Use **`create_lc_store`** or **`create_kv_docstore`** with untrusted documents.\n11. Use **`MultiVectorRetriever`** with byte stores containing untrusted documents.\n12. Use **`LangSmithRunChatLoader`** with runs containing untrusted messages.\n\nThe most common attack vector is through **LLM response fields** like `additional_kwargs` or `response_metadata`, which can be controlled via prompt injection and then serialized/deserialized in streaming operations.\n\n## Impact\n\nAttackers who control serialized data can extract environment variable secrets by injecting `{\"lc\": 1, \"type\": \"secret\", \"id\": [\"ENV_VAR\"]}` to load environment variables during deserialization (when `secrets_from_env=True`, which was the old default). They can also instantiate classes with controlled parameters by injecting constructor structures to instantiate any class within trusted namespaces with attacker-controlled parameters, potentially triggering side effects such as network calls or file operations.\n\nKey severity factors:\n\n- Affects the serialization path - applications trusting their own serialization output are vulnerable\n- Enables secret extraction when combined with `secrets_from_env=True` (the old default)\n- LLM responses in `additional_kwargs` can be controlled via prompt injection\n\n## Exploit example\n\n```python\nfrom langchain_core.load import dumps, load\nimport os\n\n# Attacker injects secret structure into user-controlled data\nattacker_dict = {\n    \"user_data\": {\n        \"lc\": 1,\n        \"type\": \"secret\",\n        \"id\": [\"OPENAI_API_KEY\"]\n    }\n}\n\nserialized = dumps(attacker_dict)  # Bug: does NOT escape the \u0027lc\u0027 key\n\nos.environ[\"OPENAI_API_KEY\"] = \"sk-secret-key-12345\"\ndeserialized = load(serialized, secrets_from_env=True)\n\nprint(deserialized[\"user_data\"])  # \"sk-secret-key-12345\" - SECRET LEAKED!\n\n```\n\n## Security hardening changes (breaking changes)\n\nThis patch introduces three breaking changes to `load()` and `loads()`:\n\n1. **New `allowed_objects` parameter** (defaults to `\u0027core\u0027`): Enforces allowlist of classes that can be deserialized. The `\u0027all\u0027` option corresponds to the list of objects [specified in `mappings.py`](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/load/mapping.py) while the `\u0027core\u0027` option limits to objects within `langchain_core`. We recommend that users explicitly specify which objects they want to allow for serialization/deserialization.\n2. **`secrets_from_env` default changed from `True` to `False`**: Disables automatic secret loading from environment\n3. **New `init_validator` parameter** (defaults to `default_init_validator`): Blocks Jinja2 templates by default\n\n## Migration guide\n\n### No changes needed for most users\n\nIf you\u0027re deserializing standard LangChain types (messages, documents, prompts, trusted partner integrations like `ChatOpenAI`, `ChatAnthropic`, etc.), your code will work without changes:\n\n```python\nfrom langchain_core.load import load\n\n# Uses default allowlist from serialization mappings\nobj = load(serialized_data)\n\n```\n\n### For custom classes\n\nIf you\u0027re deserializing custom classes not in the serialization mappings, add them to the allowlist:\n\n```python\nfrom langchain_core.load import load\nfrom my_package import MyCustomClass\n\n# Specify the classes you need\nobj = load(serialized_data, allowed_objects=[MyCustomClass])\n```\n\n### For Jinja2 templates\n\nJinja2 templates are now blocked by default because they can execute arbitrary code. If you need Jinja2 templates, pass `init_validator=None`:\n\n```python\nfrom langchain_core.load import load\nfrom langchain_core.prompts import PromptTemplate\n\nobj = load(\n    serialized_data,\n    allowed_objects=[PromptTemplate],\n    init_validator=None\n)\n\n```\n\n\u003e [!WARNING]\n\u003e Only disable `init_validator` if you trust the serialized data. Jinja2 templates can execute arbitrary Python code.\n\n### For secrets from environment\n\n`secrets_from_env` now defaults to `False`. If you need to load secrets from environment variables:\n\n```python\nfrom langchain_core.load import load\n\nobj = load(serialized_data, secrets_from_env=True)\n```\n\n\n## Credits\n\n* Dumps bug was reported by @yardenporat\n* Changes for security hardening due to findings from @0xn3va and @VladimirEliTokarev",
  "id": "BREW-langgraph-cli-CVE-2025-68664",
  "modified": "2026-09-10T00:18:25Z",
  "published": "2026-08-13T17:02:14Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/security/advisories/GHSA-c67j-w6g6-q2cm"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2025-68664"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/pull/34455"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/pull/34458"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/commit/5ec0fa69de31bbe3d76e4cf9cd65a6accb8466c8"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/commit/d9ec4c5cc78960abd37da79b0250f5642e6f0ce6"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/langchain-ai/langchain"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D0.3.81"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D1.2.5"
    }
  ],
  "schema_version": "1.7.3",
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N",
      "type": "CVSS_V3"
    }
  ],
  "summary": "LangChain serialization injection vulnerability enables secret extraction in dumps/loads APIs",
  "upstream": [
    "GHSA-c67j-w6g6-q2cm",
    "CVE-2025-68664",
    "PYSEC-2026-373"
  ]
}

{
  "$ref": "https://www.cert.ssi.gouv.fr/openapi.json",
  "affected_systems": [
    {
      "description": "WebSphere Application Server versions 8.x ant\u00e9rieures \u00e0 8.5.5.30",
      "product": {
        "name": "WebSphere",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "WebSphere Application Server - Liberty versions ant\u00e9rieures \u00e0 26.0.0.3",
      "product": {
        "name": "WebSphere",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "QRadar AI Assistant versions ant\u00e9rieures \u00e0 1.3.1",
      "product": {
        "name": "QRadar Assistant",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "Sterling Secure Proxy versions 6.1.x ant\u00e9rieures \u00e0 6.1.0.3 GA",
      "product": {
        "name": "Sterling",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "QRadar SIEM versions 7.5.x ant\u00e9rieures \u00e0 7.5.0 UP14 IF05",
      "product": {
        "name": "QRadar",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "Sterling Transformation Extender sans l\u0027application des mesures de contournement d\u00e9crites par l\u0027\u00e9diteur",
      "product": {
        "name": "Sterling",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "Sterling Secure Proxy versions 6.2.1.x ant\u00e9rieures \u00e0 6.2.1.1 GA",
      "product": {
        "name": "Sterling",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "Db2 mirror pour i sans les derniers correctifs de s\u00e9curit\u00e9",
      "product": {
        "name": "Db2",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "WebSphere Application Server versions 9.x ant\u00e9rieures \u00e0 9.0.5.27",
      "product": {
        "name": "WebSphere",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "Sterling Secure Proxy versions 6.2.x ant\u00e9rieures \u00e0 6.2.0.3 GA",
      "product": {
        "name": "Sterling",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    },
    {
      "description": "Cognos Command Center versions ant\u00e9rieures \u00e0 10.2.5 FP1 IF3",
      "product": {
        "name": "Cognos Command Center",
        "vendor": {
          "name": "IBM",
          "scada": false
        }
      }
    }
  ],
  "affected_systems_content": "IBM indique les versions correctives 9.0.5.27 et 26.0.0.3 pour WebSphere Application Server seront disponibles au cours du premier trimestre 2026. La version 8.5.5.30 sera disponible au troisi\u00e8me trimestre 2026.",
  "content": "## Solutions\n\nSe r\u00e9f\u00e9rer au bulletin de s\u00e9curit\u00e9 de l\u0027\u00e9diteur pour l\u0027obtention des correctifs (cf. section Documentation).",
  "cves": [
    {
      "name": "CVE-2026-21933",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-21933"
    },
    {
      "name": "CVE-2026-21932",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-21932"
    },
    {
      "name": "CVE-2025-12816",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-12816"
    },
    {
      "name": "CVE-2025-68973",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-68973"
    },
    {
      "name": "CVE-2025-65106",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-65106"
    },
    {
      "name": "CVE-2026-22610",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-22610"
    },
    {
      "name": "CVE-2025-66412",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66412"
    },
    {
      "name": "CVE-2025-40240",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-40240"
    },
    {
      "name": "CVE-2025-69223",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69223"
    },
    {
      "name": "CVE-2025-66035",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66035"
    },
    {
      "name": "CVE-2025-68664",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-68664"
    },
    {
      "name": "CVE-2026-22701",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-22701"
    },
    {
      "name": "CVE-2026-23745",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-23745"
    },
    {
      "name": "CVE-2026-22690",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-22690"
    },
    {
      "name": "CVE-2025-15284",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-15284"
    },
    {
      "name": "CVE-2025-69230",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69230"
    },
    {
      "name": "CVE-2025-66019",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66019"
    },
    {
      "name": "CVE-2026-21925",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-21925"
    },
    {
      "name": "CVE-2025-66031",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66031"
    },
    {
      "name": "CVE-2025-69225",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69225"
    },
    {
      "name": "CVE-2026-21860",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-21860"
    },
    {
      "name": "CVE-2025-40277",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-40277"
    },
    {
      "name": "CVE-2023-53673",
      "url": "https://www.cve.org/CVERecord?id=CVE-2023-53673"
    },
    {
      "name": "CVE-2026-1615",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-1615"
    },
    {
      "name": "CVE-2025-69227",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69227"
    },
    {
      "name": "CVE-2026-1188",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-1188"
    },
    {
      "name": "CVE-2025-66471",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66471"
    },
    {
      "name": "CVE-2025-68146",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-68146"
    },
    {
      "name": "CVE-2025-66030",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66030"
    },
    {
      "name": "CVE-2025-61140",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-61140"
    },
    {
      "name": "CVE-2025-66221",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66221"
    },
    {
      "name": "CVE-2025-69228",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69228"
    },
    {
      "name": "CVE-2025-39993",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-39993"
    },
    {
      "name": "CVE-2024-29371",
      "url": "https://www.cve.org/CVERecord?id=CVE-2024-29371"
    },
    {
      "name": "CVE-2025-40154",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-40154"
    },
    {
      "name": "CVE-2025-13601",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-13601"
    },
    {
      "name": "CVE-2025-69226",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69226"
    },
    {
      "name": "CVE-2026-21945",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-21945"
    },
    {
      "name": "CVE-2025-40248",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-40248"
    },
    {
      "name": "CVE-2025-9230",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-9230"
    },
    {
      "name": "CVE-2025-69224",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69224"
    },
    {
      "name": "CVE-2025-64756",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-64756"
    },
    {
      "name": "CVE-2025-69229",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-69229"
    },
    {
      "name": "CVE-2025-68480",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-68480"
    },
    {
      "name": "CVE-2025-14847",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-14847"
    },
    {
      "name": "CVE-2025-68285",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-68285"
    },
    {
      "name": "CVE-2025-68615",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-68615"
    },
    {
      "name": "CVE-2026-22691",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-22691"
    },
    {
      "name": "CVE-2025-66418",
      "url": "https://www.cve.org/CVERecord?id=CVE-2025-66418"
    }
  ],
  "initial_release_date": "2026-02-27T00:00:00",
  "last_revision_date": "2026-02-27T00:00:00",
  "links": [],
  "reference": "CERTFR-2026-AVI-0224",
  "revisions": [
    {
      "description": "Version initiale",
      "revision_date": "2026-02-27T00:00:00.000000"
    }
  ],
  "risks": [
    {
      "description": "D\u00e9ni de service \u00e0 distance"
    },
    {
      "description": "Injection de code indirecte \u00e0 distance (XSS)"
    },
    {
      "description": "Ex\u00e9cution de code arbitraire \u00e0 distance"
    },
    {
      "description": "Atteinte \u00e0 l\u0027int\u00e9grit\u00e9 des donn\u00e9es"
    },
    {
      "description": "Contournement de la politique de s\u00e9curit\u00e9"
    },
    {
      "description": "Atteinte \u00e0 la confidentialit\u00e9 des donn\u00e9es"
    }
  ],
  "summary": "De multiples vuln\u00e9rabilit\u00e9s ont \u00e9t\u00e9 d\u00e9couvertes dans les produits IBM. Certaines d\u0027entre elles permettent \u00e0 un attaquant de provoquer une ex\u00e9cution de code arbitraire \u00e0 distance, un d\u00e9ni de service \u00e0 distance et une atteinte \u00e0 la confidentialit\u00e9 des donn\u00e9es.",
  "title": "Multiples vuln\u00e9rabilit\u00e9s dans les produits IBM",
  "vendor_advisories": [
    {
      "published_at": "2026-02-26",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261959",
      "url": "https://www.ibm.com/support/pages/node/7261959"
    },
    {
      "published_at": "2026-02-25",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261794",
      "url": "https://www.ibm.com/support/pages/node/7261794"
    },
    {
      "published_at": "2026-02-25",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261890",
      "url": "https://www.ibm.com/support/pages/node/7261890"
    },
    {
      "published_at": "2026-02-25",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261887",
      "url": "https://www.ibm.com/support/pages/node/7261887"
    },
    {
      "published_at": "2026-02-26",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261935",
      "url": "https://www.ibm.com/support/pages/node/7261935"
    },
    {
      "published_at": "2026-02-20",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261436",
      "url": "https://www.ibm.com/support/pages/node/7261436"
    },
    {
      "published_at": "2026-02-25",
      "title": "Bulletin de s\u00e9curit\u00e9 IBM 7261774",
      "url": "https://www.ibm.com/support/pages/node/7261774"
    }
  ]
}

{
  "affected": [
    {
      "affectedData": [
        {
          "product": "langchain",
          "vendor": "langchain-ai",
          "versions": [
            {
              "status": "affected",
              "version": "\u003e= 1.0.0, \u003c 1.2.5"
            },
            {
              "status": "affected",
              "version": "\u003c 0.3.81"
            }
          ]
        }
      ],
      "source": "security-advisories@github.com"
    }
  ],
  "configurations": [
    {
      "nodes": [
        {
          "cpeMatch": [
            {
              "criteria": "cpe:2.3:a:langchain:langchain_core:*:*:*:*:*:python:*:*",
              "matchCriteriaId": "DC184324-6CF1-4F7A-B87F-6DD2120C3B3B",
              "versionEndExcluding": "0.3.81",
              "vulnerable": true
            },
            {
              "criteria": "cpe:2.3:a:langchain:langchain_core:*:*:*:*:*:python:*:*",
              "matchCriteriaId": "4298E953-ED96-49EC-8474-86095D560F2B",
              "versionEndExcluding": "1.2.5",
              "versionStartIncluding": "1.0.0",
              "vulnerable": true
            }
          ],
          "negate": false,
          "operator": "OR"
        }
      ]
    }
  ],
  "cveTags": [],
  "descriptions": [
    {
      "lang": "en",
      "value": "LangChain is a framework for building agents and LLM-powered applications. Prior to versions 0.3.81 and 1.2.5, a serialization injection vulnerability exists in LangChain\u0027s dumps() and dumpd() functions. The functions do not escape dictionaries with \u0027lc\u0027 keys when serializing free-form dictionaries. The \u0027lc\u0027 key is used internally by LangChain to mark serialized objects. When user-controlled data contains this key structure, it is treated as a legitimate LangChain object during deserialization rather than plain user data. This issue has been patched in versions 0.3.81 and 1.2.5."
    },
    {
      "lang": "es",
      "value": "LangChain es un framework para construir agentes y aplicaciones impulsadas por LLM. Versiones anteriores a 0.3.81 y 1.2.5, una vulnerabilidad de inyecci\u00f3n de serializaci\u00f3n existe en las funciones dumps() y dumpd() de LangChain. Las funciones no escapan los diccionarios con claves \u2019lc\u2019 al serializar diccionarios de formato libre. La clave \u2019lc\u2019 se usa internamente en LangChain para marcar objetos serializados. Cuando datos controlados por el usuario contienen esta estructura de clave, se trata como un objeto leg\u00edtimo de LangChain durante la deserializaci\u00f3n en lugar de datos de usuario simples. Este problema ha sido corregido en las versiones 0.3.81 y 1.2.5."
    }
  ],
  "id": "CVE-2025-68664",
  "lastModified": "2026-10-19T22:10:00.130",
  "metrics": {
    "cvssMetricV31": [
      {
        "cvssData": {
          "attackComplexity": "LOW",
          "attackVector": "NETWORK",
          "availabilityImpact": "NONE",
          "baseScore": 9.3,
          "baseSeverity": "CRITICAL",
          "confidentialityImpact": "HIGH",
          "integrityImpact": "LOW",
          "privilegesRequired": "NONE",
          "scope": "CHANGED",
          "userInteraction": "NONE",
          "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N",
          "version": "3.1"
        },
        "exploitabilityScore": 3.9,
        "impactScore": 4.7,
        "source": "security-advisories@github.com",
        "type": "Secondary"
      },
      {
        "cvssData": {
          "attackComplexity": "LOW",
          "attackVector": "NETWORK",
          "availabilityImpact": "NONE",
          "baseScore": 8.2,
          "baseSeverity": "HIGH",
          "confidentialityImpact": "HIGH",
          "integrityImpact": "LOW",
          "privilegesRequired": "NONE",
          "scope": "UNCHANGED",
          "userInteraction": "NONE",
          "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:L/A:N",
          "version": "3.1"
        },
        "exploitabilityScore": 3.9,
        "impactScore": 4.2,
        "source": "nvd@nist.gov",
        "type": "Primary"
      }
    ],
    "ssvcV203": [
      {
        "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
        "ssvcData": {
          "id": "CVE-2025-68664",
          "options": [
            {
              "exploitation": "poc"
            },
            {
              "automatable": "yes"
            },
            {
              "technicalImpact": "partial"
            }
          ],
          "role": "CISA Coordinator",
          "timestamp": "2025-12-24T14:40:55.506059Z",
          "version": "2.0.3"
        }
      }
    ]
  },
  "published": "2025-12-23T23:15:44.933",
  "references": [
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Patch"
      ],
      "url": "https://github.com/langchain-ai/langchain/commit/5ec0fa69de31bbe3d76e4cf9cd65a6accb8466c8"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Patch"
      ],
      "url": "https://github.com/langchain-ai/langchain/commit/d9ec4c5cc78960abd37da79b0250f5642e6f0ce6"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Issue Tracking",
        "Patch"
      ],
      "url": "https://github.com/langchain-ai/langchain/pull/34455"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Issue Tracking",
        "Patch"
      ],
      "url": "https://github.com/langchain-ai/langchain/pull/34458"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Release Notes"
      ],
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D0.3.81"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Release Notes"
      ],
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D1.2.5"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Exploit",
        "Vendor Advisory"
      ],
      "url": "https://github.com/langchain-ai/langchain/security/advisories/GHSA-c67j-w6g6-q2cm"
    },
    {
      "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
      "tags": [
        "Exploit",
        "Vendor Advisory"
      ],
      "url": "https://github.com/langchain-ai/langchain/security/advisories/GHSA-c67j-w6g6-q2cm"
    }
  ],
  "sourceIdentifier": "security-advisories@github.com",
  "vulnStatus": "Analyzed",
  "weaknesses": [
    {
      "description": [
        {
          "lang": "en",
          "value": "CWE-502"
        }
      ],
      "source": "security-advisories@github.com",
      "type": "Secondary"
    }
  ]
}

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "langchain-core"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "1.0.0"
            },
            {
              "fixed": "1.2.5"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    },
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "langchain-core"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "0.3.81"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2025-68664"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-502"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2025-12-23T18:46:13Z",
    "nvd_published_at": "2025-12-23T23:15:44Z",
    "severity": "CRITICAL"
  },
  "details": "## Summary\n\nA serialization injection vulnerability exists in LangChain\u0027s `dumps()` and `dumpd()` functions. The functions do not escape dictionaries with `\u0027lc\u0027` keys when serializing free-form dictionaries. The `\u0027lc\u0027` key is used internally by LangChain to mark serialized objects. When user-controlled data contains this key structure, it is treated as a legitimate LangChain object during deserialization rather than plain user data.\n\n### Attack surface\n\nThe core vulnerability was in `dumps()` and `dumpd()`: these functions failed to escape user-controlled dictionaries containing `\u0027lc\u0027` keys. When this unescaped data was later deserialized via `load()` or `loads()`, the injected structures were treated as legitimate LangChain objects rather than plain user data.\n\nThis escaping bug enabled several attack vectors:\n\n1. **Injection via user data**: Malicious LangChain object structures could be injected through user-controlled fields like `metadata`, `additional_kwargs`, or `response_metadata`\n2. **Class instantiation within trusted namespaces**: Injected manifests could instantiate any `Serializable` subclass, but only within the pre-approved trusted namespaces (`langchain_core`, `langchain`, `langchain_community`). This includes classes with side effects in `__init__` (network calls, file operations, etc.). Note that namespace validation was already enforced before this patch, so arbitrary classes outside these trusted namespaces could not be instantiated.\n\n### Security hardening\n\nThis patch fixes the escaping bug in `dumps()` and `dumpd()` and introduces new restrictive defaults in `load()` and `loads()`: allowlist enforcement via `allowed_objects=\"core\"` (restricted to [serialization mappings](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/load/mapping.py)), `secrets_from_env` changed from `True` to `False`, and default Jinja2 template blocking via `init_validator`. These are breaking changes for some use cases.\n\n## Who is affected?\n\nApplications are vulnerable if they:\n\n1. **Use `astream_events(version=\"v1\")`** \u2014 The v1 implementation internally uses vulnerable serialization. Note: `astream_events(version=\"v2\")` is not vulnerable.\n2. **Use `Runnable.astream_log()`** \u2014 This method internally uses vulnerable serialization for streaming outputs.\n3. **Call `dumps()` or `dumpd()` on untrusted data, then deserialize with `load()` or `loads()`** \u2014 Trusting your own serialization output makes you vulnerable if user-controlled data (e.g., from LLM responses, metadata fields, or user inputs) contains `\u0027lc\u0027` key structures.\n4. **Deserialize untrusted data with `load()` or `loads()`** \u2014 Directly deserializing untrusted data that may contain injected `\u0027lc\u0027` structures.\n5. **Use `RunnableWithMessageHistory`** \u2014 Internal serialization in message history handling.\n6. **Use `InMemoryVectorStore.load()`** to deserialize untrusted documents.\n7. Load untrusted generations from cache using **`langchain-community` caches**.\n8. Load untrusted manifests from the LangChain Hub via **`hub.pull`**.\n9. Use **`StringRunEvaluatorChain`** on untrusted runs.\n10. Use **`create_lc_store`** or **`create_kv_docstore`** with untrusted documents.\n11. Use **`MultiVectorRetriever`** with byte stores containing untrusted documents.\n12. Use **`LangSmithRunChatLoader`** with runs containing untrusted messages.\n\nThe most common attack vector is through **LLM response fields** like `additional_kwargs` or `response_metadata`, which can be controlled via prompt injection and then serialized/deserialized in streaming operations.\n\n## Impact\n\nAttackers who control serialized data can extract environment variable secrets by injecting `{\"lc\": 1, \"type\": \"secret\", \"id\": [\"ENV_VAR\"]}` to load environment variables during deserialization (when `secrets_from_env=True`, which was the old default). They can also instantiate classes with controlled parameters by injecting constructor structures to instantiate any class within trusted namespaces with attacker-controlled parameters, potentially triggering side effects such as network calls or file operations.\n\nKey severity factors:\n\n- Affects the serialization path - applications trusting their own serialization output are vulnerable\n- Enables secret extraction when combined with `secrets_from_env=True` (the old default)\n- LLM responses in `additional_kwargs` can be controlled via prompt injection\n\n## Exploit example\n\n```python\nfrom langchain_core.load import dumps, load\nimport os\n\n# Attacker injects secret structure into user-controlled data\nattacker_dict = {\n    \"user_data\": {\n        \"lc\": 1,\n        \"type\": \"secret\",\n        \"id\": [\"OPENAI_API_KEY\"]\n    }\n}\n\nserialized = dumps(attacker_dict)  # Bug: does NOT escape the \u0027lc\u0027 key\n\nos.environ[\"OPENAI_API_KEY\"] = \"sk-secret-key-12345\"\ndeserialized = load(serialized, secrets_from_env=True)\n\nprint(deserialized[\"user_data\"])  # \"sk-secret-key-12345\" - SECRET LEAKED!\n\n```\n\n## Security hardening changes (breaking changes)\n\nThis patch introduces three breaking changes to `load()` and `loads()`:\n\n1. **New `allowed_objects` parameter** (defaults to `\u0027core\u0027`): Enforces allowlist of classes that can be deserialized. The `\u0027all\u0027` option corresponds to the list of objects [specified in `mappings.py`](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/load/mapping.py) while the `\u0027core\u0027` option limits to objects within `langchain_core`. We recommend that users explicitly specify which objects they want to allow for serialization/deserialization.\n2. **`secrets_from_env` default changed from `True` to `False`**: Disables automatic secret loading from environment\n3. **New `init_validator` parameter** (defaults to `default_init_validator`): Blocks Jinja2 templates by default\n\n## Migration guide\n\n### No changes needed for most users\n\nIf you\u0027re deserializing standard LangChain types (messages, documents, prompts, trusted partner integrations like `ChatOpenAI`, `ChatAnthropic`, etc.), your code will work without changes:\n\n```python\nfrom langchain_core.load import load\n\n# Uses default allowlist from serialization mappings\nobj = load(serialized_data)\n\n```\n\n### For custom classes\n\nIf you\u0027re deserializing custom classes not in the serialization mappings, add them to the allowlist:\n\n```python\nfrom langchain_core.load import load\nfrom my_package import MyCustomClass\n\n# Specify the classes you need\nobj = load(serialized_data, allowed_objects=[MyCustomClass])\n```\n\n### For Jinja2 templates\n\nJinja2 templates are now blocked by default because they can execute arbitrary code. If you need Jinja2 templates, pass `init_validator=None`:\n\n```python\nfrom langchain_core.load import load\nfrom langchain_core.prompts import PromptTemplate\n\nobj = load(\n    serialized_data,\n    allowed_objects=[PromptTemplate],\n    init_validator=None\n)\n\n```\n\n\u003e [!WARNING]\n\u003e Only disable `init_validator` if you trust the serialized data. Jinja2 templates can execute arbitrary Python code.\n\n### For secrets from environment\n\n`secrets_from_env` now defaults to `False`. If you need to load secrets from environment variables:\n\n```python\nfrom langchain_core.load import load\n\nobj = load(serialized_data, secrets_from_env=True)\n```\n\n\n## Credits\n\n* Dumps bug was reported by @yardenporat\n* Changes for security hardening due to findings from @0xn3va and @VladimirEliTokarev",
  "id": "GHSA-c67j-w6g6-q2cm",
  "modified": "2026-10-27T16:50:48Z",
  "published": "2025-12-23T18:46:13Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/security/advisories/GHSA-c67j-w6g6-q2cm"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2025-68664"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/pull/34455"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/pull/34458"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/commit/5ec0fa69de31bbe3d76e4cf9cd65a6accb8466c8"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/commit/d9ec4c5cc78960abd37da79b0250f5642e6f0ce6"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/langchain-ai/langchain"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D0.3.81"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D1.2.5"
    },
    {
      "type": "WEB",
      "url": "https://github.com/pypa/advisory-database/tree/main/vulns/langchain-core/PYSEC-2026-373.yaml"
    },
    {
      "type": "WEB",
      "url": "https://pypi.org/project/langchain-core"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N",
      "type": "CVSS_V3"
    }
  ],
  "summary": "LangChain serialization injection vulnerability enables secret extraction in dumps/loads APIs"
}

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "langchain-core",
        "purl": "pkg:pypi/langchain-core"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "0.3.81"
            },
            {
              "introduced": "1.0.0"
            },
            {
              "fixed": "1.2.5"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ],
      "versions": [
        "0.0.1",
        "0.0.10",
        "0.0.11",
        "0.0.12",
        "0.0.13",
        "0.0.13rc1",
        "0.0.13rc2",
        "0.0.2",
        "0.0.3",
        "0.0.4",
        "0.0.5",
        "0.0.6",
        "0.0.7",
        "0.0.8",
        "0.0.9",
        "0.1.0",
        "0.1.1",
        "0.1.10",
        "0.1.11",
        "0.1.12",
        "0.1.12rc1",
        "0.1.13",
        "0.1.14",
        "0.1.15",
        "0.1.15rc1",
        "0.1.15rc2",
        "0.1.16",
        "0.1.17",
        "0.1.18",
        "0.1.19",
        "0.1.2",
        "0.1.20",
        "0.1.21",
        "0.1.22",
        "0.1.23",
        "0.1.24",
        "0.1.25",
        "0.1.26",
        "0.1.27",
        "0.1.28",
        "0.1.29",
        "0.1.3",
        "0.1.30",
        "0.1.31",
        "0.1.32",
        "0.1.33",
        "0.1.33rc1",
        "0.1.34",
        "0.1.35",
        "0.1.36",
        "0.1.37",
        "0.1.38",
        "0.1.39",
        "0.1.4",
        "0.1.40",
        "0.1.41",
        "0.1.42",
        "0.1.42rc1",
        "0.1.43",
        "0.1.44",
        "0.1.45",
        "0.1.46",
        "0.1.47",
        "0.1.47rc1",
        "0.1.48",
        "0.1.49",
        "0.1.5",
        "0.1.50",
        "0.1.51",
        "0.1.52",
        "0.1.53",
        "0.1.6",
        "0.1.7",
        "0.1.8",
        "0.1.9",
        "0.2.0",
        "0.2.0rc1",
        "0.2.1",
        "0.2.10",
        "0.2.11",
        "0.2.12",
        "0.2.13",
        "0.2.15",
        "0.2.16",
        "0.2.17",
        "0.2.18",
        "0.2.19",
        "0.2.2",
        "0.2.20",
        "0.2.21",
        "0.2.22",
        "0.2.23",
        "0.2.24",
        "0.2.25",
        "0.2.26",
        "0.2.27",
        "0.2.28",
        "0.2.29",
        "0.2.29rc1",
        "0.2.2rc1",
        "0.2.3",
        "0.2.30",
        "0.2.31",
        "0.2.32",
        "0.2.33",
        "0.2.34",
        "0.2.35",
        "0.2.36",
        "0.2.37",
        "0.2.38",
        "0.2.39",
        "0.2.4",
        "0.2.40",
        "0.2.41",
        "0.2.42",
        "0.2.43",
        "0.2.5",
        "0.2.6",
        "0.2.7",
        "0.2.8",
        "0.2.9",
        "0.3.0",
        "0.3.0.dev0",
        "0.3.0.dev1",
        "0.3.0.dev2",
        "0.3.0.dev3",
        "0.3.0.dev4",
        "0.3.0.dev5",
        "0.3.1",
        "0.3.10",
        "0.3.11",
        "0.3.12",
        "0.3.13",
        "0.3.14",
        "0.3.15",
        "0.3.16",
        "0.3.17",
        "0.3.18",
        "0.3.19",
        "0.3.2",
        "0.3.20",
        "0.3.21",
        "0.3.22",
        "0.3.23",
        "0.3.24",
        "0.3.25",
        "0.3.26",
        "0.3.27",
        "0.3.28",
        "0.3.29",
        "0.3.3",
        "0.3.30",
        "0.3.31",
        "0.3.32",
        "0.3.33",
        "0.3.34",
        "0.3.34rc1",
        "0.3.34rc2",
        "0.3.35",
        "0.3.36",
        "0.3.37",
        "0.3.38",
        "0.3.39",
        "0.3.4",
        "0.3.40",
        "0.3.41",
        "0.3.42",
        "0.3.43",
        "0.3.44",
        "0.3.45",
        "0.3.45rc1",
        "0.3.46",
        "0.3.47",
        "0.3.48",
        "0.3.49",
        "0.3.5",
        "0.3.50",
        "0.3.51",
        "0.3.52",
        "0.3.53",
        "0.3.54",
        "0.3.55",
        "0.3.56",
        "0.3.56rc1",
        "0.3.57",
        "0.3.58",
        "0.3.59",
        "0.3.6",
        "0.3.60",
        "0.3.61",
        "0.3.62",
        "0.3.63",
        "0.3.64",
        "0.3.65",
        "0.3.66",
        "0.3.67",
        "0.3.68",
        "0.3.69",
        "0.3.7",
        "0.3.70",
        "0.3.71",
        "0.3.72",
        "0.3.73",
        "0.3.74",
        "0.3.75",
        "0.3.76",
        "0.3.77",
        "0.3.78",
        "0.3.79",
        "0.3.8",
        "0.3.80",
        "0.3.9",
        "1.0.0",
        "1.0.1",
        "1.0.2",
        "1.0.3",
        "1.0.4",
        "1.0.5",
        "1.0.6",
        "1.0.7",
        "1.1.0",
        "1.1.1",
        "1.1.2",
        "1.1.3",
        "1.2.0",
        "1.2.1",
        "1.2.2",
        "1.2.3",
        "1.2.4"
      ]
    }
  ],
  "aliases": [
    "CVE-2025-68664",
    "GHSA-c67j-w6g6-q2cm"
  ],
  "details": "## Summary\n\nA serialization injection vulnerability exists in LangChain\u0027s `dumps()` and `dumpd()` functions. The functions do not escape dictionaries with `\u0027lc\u0027` keys when serializing free-form dictionaries. The `\u0027lc\u0027` key is used internally by LangChain to mark serialized objects. When user-controlled data contains this key structure, it is treated as a legitimate LangChain object during deserialization rather than plain user data.\n\n### Attack surface\n\nThe core vulnerability was in `dumps()` and `dumpd()`: these functions failed to escape user-controlled dictionaries containing `\u0027lc\u0027` keys. When this unescaped data was later deserialized via `load()` or `loads()`, the injected structures were treated as legitimate LangChain objects rather than plain user data.\n\nThis escaping bug enabled several attack vectors:\n \n1. **Injection via user data**: Malicious LangChain object structures could be injected through user-controlled fields like `metadata`, `additional_kwargs`, or `response_metadata`\n2. **Class instantiation within trusted namespaces**: Injected manifests could instantiate any `Serializable` subclass, but only within the pre-approved trusted namespaces (`langchain_core`, `langchain`, `langchain_community`). This includes classes with side effects in `__init__` (network calls, file operations, etc.). Note that namespace validation was already enforced before this patch, so arbitrary classes outside these trusted namespaces could not be instantiated.\n\n ### Security hardening\n\nThis patch fixes the escaping bug in `dumps()` and `dumpd()` and introduces new restrictive defaults in `load()` and `loads()`: allowlist enforcement via `allowed_objects=\"core\"` (restricted to [serialization mappings](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/load/mapping.py)), `secrets_from_env` changed from `True` to `False`, and default Jinja2 template blocking via `init_validator`. These are breaking changes for some use cases.\n\n## Who is affected?\n\nApplications are vulnerable if they:\n\n1. **Use `astream_events(version=\"v1\")`** \u2014 The v1 implementation internally uses vulnerable serialization. Note: `astream_events(version=\"v2\")` is not vulnerable.\n2. **Use `Runnable.astream_log()`** \u2014 This method internally uses vulnerable serialization for streaming outputs.\n 3. **Call `dumps()` or `dumpd()` on untrusted data, then deserialize with `load()` or `loads()`** \u2014 Trusting your own serialization output makes you vulnerable if user-controlled data (e.g., from LLM responses, metadata fields, or user inputs) contains `\u0027lc\u0027` key structures.\n4. **Deserialize untrusted data with `load()` or `loads()`** \u2014 Directly deserializing untrusted data that may contain injected `\u0027lc\u0027` structures.\n5. **Use `RunnableWithMessageHistory`** \u2014 Internal serialization in message history handling.\n6. **Use `InMemoryVectorStore.load()`** to deserialize untrusted documents.\n7. Load untrusted generations from cache using **`langchain-community` caches**.\n8. Load untrusted manifests from the LangChain Hub via **`hub.pull`**.\n 9. Use **`StringRunEvaluatorChain`** on untrusted runs.\n10. Use **`create_lc_store`** or **`create_kv_docstore`** with untrusted documents.\n11. Use **`MultiVectorRetriever`** with byte stores containing untrusted documents.\n12. Use **`LangSmithRunChatLoader`** with runs containing untrusted messages.\n\nThe most common attack vector is through **LLM response fields** like `additional_kwargs` or `response_metadata`, which can be controlled via prompt injection and then serialized/deserialized in streaming operations.\n\n## Impact\n\nAttackers who control serialized data can extract environment variable secrets by injecting `{\"lc\": 1, \"type\": \"secret\", \"id\": [\"ENV_VAR\"]}` to load environment variables during deserialization (when `secrets_from_env=True`, which was the old default). They can also instantiate classes with controlled parameters by injecting constructor structures to instantiate any class within trusted namespaces with attacker-controlled parameters, potentially triggering side effects such as network calls or file operations.\n\nKey severity factors:\n\n- Affects the serialization path - applications trusting their own serialization output are vulnerable\n- Enables secret extraction when combined with `secrets_from_env=True` (the old default)\n - LLM responses in `additional_kwargs` can be controlled via prompt injection\n\n ## Exploit example\n\n```python\nfrom langchain_core.load import dumps, load\nimport os\n\n# Attacker injects secret structure into user-controlled data\nattacker_dict = {\n    \"user_data\": {\n        \"lc\": 1,\n        \"type\": \"secret\",\n        \"id\": [\"OPENAI_API_KEY\"]\n    }\n}\n\nserialized = dumps(attacker_dict)  # Bug: does NOT escape the \u0027lc\u0027 key\n\nos.environ[\"OPENAI_API_KEY\"] = \"sk-secret-key-12345\"\ndeserialized = load(serialized, secrets_from_env=True)\n\nprint(deserialized[\"user_data\"])  # \"sk-secret-key-12345\" - SECRET LEAKED!\n\n```\n\n## Security hardening changes (breaking changes)\n\nThis patch introduces three breaking changes to `load()` and `loads()`:\n\n1. **New `allowed_objects` parameter** (defaults to `\u0027core\u0027`): Enforces allowlist of classes that can be deserialized. The `\u0027all\u0027` option corresponds to the list of objects [specified in `mappings.py`](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/load/mapping.py) while the `\u0027core\u0027` option limits to objects within `langchain_core`. We recommend that users explicitly specify which objects they want to allow for serialization/deserialization.\n 2. **`secrets_from_env` default changed from `True` to `False`**: Disables automatic secret loading from environment\n3. **New `init_validator` parameter** (defaults to `default_init_validator`): Blocks Jinja2 templates by default\n\n## Migration guide\n\n### No changes needed for most users\n\nIf you\u0027re deserializing standard LangChain types (messages, documents, prompts, trusted partner integrations like `ChatOpenAI`, `ChatAnthropic`, etc.), your code will work without changes:\n\n```python\n from langchain_core.load import load\n\n# Uses default allowlist from serialization mappings\nobj = load(serialized_data)\n\n```\n\n### For custom classes\n\nIf you\u0027re deserializing custom classes not in the serialization mappings, add them to the allowlist:\n\n```python\nfrom langchain_core.load import load\nfrom my_package import MyCustomClass\n\n# Specify the classes you need\nobj = load(serialized_data, allowed_objects=[MyCustomClass])\n ```\n\n### For Jinja2 templates\n\nJinja2 templates are now blocked by default because they can execute arbitrary code. If you need Jinja2 templates, pass `init_validator=None`:\n \n```python\nfrom langchain_core.load import load\nfrom langchain_core.prompts import PromptTemplate\n\nobj = load(\n    serialized_data,\n    allowed_objects=[PromptTemplate],\n    init_validator=None\n)\n\n```\n\n\u003e [!WARNING]\n\u003e Only disable `init_validator` if you trust the serialized data. Jinja2 templates can execute arbitrary Python code.\n\n### For secrets from environment\n\n`secrets_from_env` now defaults to `False`. If you need to load secrets from environment variables:\n\n```python\n from langchain_core.load import load\n\nobj = load(serialized_data, secrets_from_env=True)\n ```\n\n\n## Credits\n\n* Dumps bug was reported by @yardenporat\n* Changes for security hardening due to findings from @0xn3va and @VladimirEliTokarev",
  "id": "PYSEC-2026-373",
  "modified": "2026-07-02T12:46:34.720444Z",
  "published": "2026-06-29T11:50:38.732432Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/security/advisories/GHSA-c67j-w6g6-q2cm"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2025-68664"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/pull/34455"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/pull/34458"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/commit/5ec0fa69de31bbe3d76e4cf9cd65a6accb8466c8"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/commit/d9ec4c5cc78960abd37da79b0250f5642e6f0ce6"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/langchain-ai/langchain"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D0.3.81"
    },
    {
      "type": "WEB",
      "url": "https://github.com/langchain-ai/langchain/releases/tag/langchain-core%3D%3D1.2.5"
    },
    {
      "type": "PACKAGE",
      "url": "https://pypi.org/project/langchain-core"
    },
    {
      "type": "ADVISORY",
      "url": "https://github.com/advisories/GHSA-c67j-w6g6-q2cm"
    }
  ],
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N",
      "type": "CVSS_V3"
    }
  ],
  "summary": "LangChain serialization injection vulnerability enables secret extraction in dumps/loads APIs"
}