Anthropic AI wellbeing research grants
Anthropic offers $5 million in grants for open source evaluations of how AI affects user wellbeing. First applications are due by September 21, 2026.
Anthropic has launched a $5 million grant program for independent research into how AI affects the wellbeing of the people who use it. Grantees get direct funding, access to Anthropic models and technical support. The goal is open source evaluations that measure this impact.
The program is aimed at people with expertise in the field. Anthropic names clinicians, psychologists and methodologists, and welcomes others with relevant experience. The evaluations should cover situations such as emotional support and mental health crises.
Applications go through a short form on the Anthropic website. Applications are due by September 21, 2026. Anthropic will tell selected applicants by October 5 whether they may submit a full proposal.
Anthropic says it wants evaluations that state clearly what they measure and what counts as a pass or a fail. Clinical and subject matter experts should be involved in the design and validation. Evaluations should test both harms and precautions, so that they catch too much refusal as well as too little. They should reflect how people actually use AI, which often means conversations with many turns. Graders should be validated against real experts.