An update on our mental health-related work

Source: OpenAI Date: 2026-02-27 URL: https://openai.com/index/update-on-mental-health-related-work

Summary

OpenAI’s update on their research and product work related to mental health — covering how ChatGPT handles mental health conversations, what safeguards exist for users in crisis, what research they’re doing on AI’s effects on emotional well-being, and their partnerships with mental health organizations. Published alongside the Expert Council on Well-being and AI announcement (October 2025) and teen safety blueprint.

Implications

Safety/well-being thread. Mental health is one of the most sensitive domains for AI deployment: the risk of harm (reinforcing negative thinking, missing suicidal ideation, failing to refer to professionals) is high, and the potential benefit (accessible mental health support for people without access to care) is also high. OpenAI’s February 2026 update signals ongoing investment in this domain rather than a one-time policy statement. The update likely responds to academic research on AI chatbot effects on mental health outcomes that emerged in 2025.