Trust but Verify: Mitigating Medical Hallucinations via Post-Hoc Adversarial Auditing and Multi-Agent Feedback Loops

In three linesarXiv study on medical hallucinations in LLMs. "Trust but Verify" multi-agent system testing 3 model families (GPT-OSS, Llama-3, Falcon-3) on 103 clinical MCQs featuring banned pharmaceuticals. Five-agent architecture reduces Hallucination Error Rate by 53% and enforces appropriate refusal over dangerous recommendations.

Summary generated by Claude — human-verified