OpenAI announced the launch of MentalHealthBench, an open benchmark developed in collaboration with over 80 licensed mental health experts from 22 countries. The benchmark evaluates AI systems' responses in realistic mental health scenarios, focusing on key behaviors such as safety, context-seeking, user agency preservation, and providing actionable guidance. According to the announcement, MentalHealthBench aims to set a higher standard for AI support in mental health contexts, helping researchers identify gaps in existing models and improve future iterations. OpenAI also highlighted its efforts to enhance ChatGPT's responses in sensitive conversations, expand access to crisis resources, and introduce features like Trusted Contact for moments of distress. Additionally, the company is supporting new research initiatives and collaborating with the Partnership on AI to advance AI applications in mental health.
Source: OpenAI
