Funding grants for new research into AI and mental health

Source: OpenAI Date: 2025-12-01 URL: https://openai.com/index/ai-mental-health-research-grants

Summary

OpenAI grant program announcement from December 2025 funding academic and clinical research into AI’s effects on mental health — both potential harms (AI relationship dependency, impact on adolescent mental health, displacement effects) and potential benefits (AI-assisted therapy tools, mental health support accessibility). The timing aligns with the teen safety model spec update and the NORAD partnership — a cluster of December 2025 announcements addressing social responsibility concerns about AI.

Implications

Preemptive research funding as liability management. Funding external research into AI’s mental health impacts is both genuinely useful (the research is needed) and strategically defensive — if OpenAI funds the research, they have some input into the framing and access to results before publication. Meta faced significant backlash when internal research on teen mental health leaked without context; OpenAI is building an external research ecosystem instead.

The AI companion question. One of the most significant mental health concerns about ChatGPT and similar products is the risk of users forming unhealthy emotional dependencies — using AI as a substitute for human relationships in ways that may increase isolation. Whether the funded research investigates this question honestly is the integrity test.

Thread: consumer AI safety governance. Sits alongside the Thrive Holdings ownership stake (December 2025), the teen protections model spec update, and the SafetyKit deployment as markers of OpenAI’s late-2025 consumer safety investment cluster.

Watch: Whether the funded research produces results that inform OpenAI product decisions in ways that create user-protective product changes, or whether the research serves primarily as a PR buffer without product-side impact.