This article discusses a new paper outlining design patterns for mitigating prompt injection attacks in LLM agents. It details six patterns – Action-Selector, Plan-Then-Execute, LLM Map-Reduce, Dual LLM, Code-Then-Execute, and Context-Minimization – and emphasizes the need for trade-offs between agent utility and security by limiting the ability of agents to perform arbitrary tasks.
Simon Willison explains an accidental prompt injection attack on RAG applications, caused by concatenating user questions with documentation fragments in a Retrieval Augmented Generation (RAG) system.
