LangChain, LangGraph Flaws Expose Files, Secrets, Databases in Widely Used AI Frameworks

Three security vulnerabilities in LangChain and LangGraph frameworks expose filesystem data, environment secrets, and conversation history in enterprise AI deployments.

• •CVE-2026-34070 (CVSS 7.5): Path traversal in langchain_core/prompts/loading.py allows arbitrary file access via crafted prompt templates without validation.
• •CVE-2025-68664 (CVSS 9.3): Deserialization vulnerability leaks API keys and environment secrets by tricking the app into treating user input as a serialized LangChain object.
• •CVE-2025-67644 (CVSS 7.3): SQL injection in LangGraph's SQLite checkpoint implementation allows arbitrary SQL queries via manipulated metadata filter keys.
• •Patches are available: langchain-core >=1.2.22, langchain-core 0.3.81/1.2.5, and langgraph-checkpoint-sqlite 3.0.1.
• •