Known CVEpypi/langchain-core
CVE-2025-68664: LangChain serialization injection exposes secrets (LangGrinch)
Critical
dumps/loads do not escape user-controlled dicts containing the reserved 'lc' key, so attacker-shaped data (for example from LLM output or metadata) is revived as LangChain objects on load, which can read secrets from environment variables. Fixed in 0.3.81 and 1.2.5.
Affected and fixed versions
Affected: < 0.3.81; >= 1.0.0, < 1.2.5. Fixed in 0.3.81, 1.2.5.
Sources
See every known CVE, and the ones in the news, in the Learning Center