The path traversal flaw, allowing access to arbitrary files, adds to a growing set of input validation issues in AI pipelines.
Three LangChain flaws enable data theft across LLM apps, affecting millions of deployments, exposing secrets and files.