BIPIA

A 2 October 2026 arXiv paper evaluating fifteen prompt-injection detectors, including Meta's Prompt Guard 2, found that public benchmark scores transfer poorly to real LLM agent deployments: the best …

Mozilla AI's experiments with its any-guardrail tool found that the open-source PIGuard model effectively detected indirect prompt injection attacks on the email and tabular portions of Microsoft's BI…

An independent researcher found that existing indirect prompt injection benchmarks like BIPIA, InjecAgent, and AgentDojo may produce unreliable scores due to reliance on LLM-judges and evaluation of e…