CVE-2026-44843
high
CVSS v3
8.2
CVSS v4 NEW
โ
VIR risk
8.2
Description
LangChain vulnerable to unsafe deserialization of attacker-controlled objects through overly broad `load()` allowlists
Predictions
Exploit likelihood
88%
Patch ETA
โ
Heuristic predictions, AS-IS, for prioritization only.
Mitigations
{**Immediate action:** Upgrade to patched versions:
```bash
pip install --upgrade langchain>=0.3.85 langchain-core>=0.3.85
# For 1.x users
pip install --upgrade langchain>=1.3.3
```
**Interim workaround** if upgrade is blocked: ensure `load()` is never called on untrusted input. Wrap deserialization with input validation:
```python
from langchain.load import loads
def safe_load(data: str, allowed_namespaces: list[str]):
obj = loads(data)
if not obj.get("id", [None])[0] in allowed_namespaces:
raise ValueError("Untrusted namespace")
return obj
```
Audit any custom callback handlers or tool wrappers that deserialize agent state. **Rollback:** downgrade carries the vulnerability; test upgraded version in staging with full agent workflows before production deployment.}
Package impact
| Ecosystem | Package | Vulnerable | Fixed |
|---|---|---|---|
| PyPI | langchain-core | >=1.0.0,<1.3.3 | 1.3.3 |
| PyPI | langchain-core | <0.3.85 | 0.3.85 |
| PIP | langchain-core | <= 0.3.84 | 0.3.85 |
| PIP | langchain-core | >= 1.0.0, <= 1.3.2 | 1.3.3 |
References
CWEs
CWE-502
Community-verified mitigations for this CVE will appear above when contributors publish them.
Verify integrity in audit chain (admin only). AS-IS.