#ai-llm-rag-machine-learning-vector-databases-nlp

RAG poisoning: how one document can hijack your AI agent

Indirect prompt injection is the attack your prompt filter never sees. Here's how it works — with a concrete example — and how to defend against it.