The Hallway Track
Research Findings

Introducing MentalHealthBench

OpenAI · OpenAI Blog · Sep 23, 2026 · Research Findings

OpenAI introduces MentalHealthBench, an expert-informed benchmark for evaluating safe AI responses in mental health conversations.

“MentalHealthBench is an expert-informed benchmark for evaluating helpful and safe AI responses across realistic mental health conversations.”— OpenAI

OpenAI released MentalHealthBench, an expert-informed benchmark measuring whether AI models respond helpfully and safely in realistic mental health conversations. It matters as a formal evaluation standard for high-stakes safety-critical AI use cases, signaling growing industry focus on measurable safety guardrails.

benchmark mental-health ai-safety openai evaluation

Watch / read the original source →