MedPRESS: A Multi-turn Benchmark for Patient-Pressure-Induced Medical Sycophancy in LLMs

In 30 Sekunden

Was passiert ist

Researchers released MedPRESS, a benchmark testing whether medical LLMs maintain safe advice when patients pressure them across five-turn conversations.

Warum es zählt

Engineers building or deploying medical chatbots need this to understand how models degrade under realistic patient interaction patterns.

Achtung

The benchmark tests 20 models but does not reveal which specific models fail worst, limiting direct actionability for practitioners.