Funding better evaluations of AI’s impact on wellbeing
anthropic ·
Anthropic is launching a $5 million grant program to fund independent research on how AI affects user wellbeing, offering funding, model access, and technical support. Grantees will create open-source evaluations that any developer can use to measure AI's impact. The company notes that AI systems are now used for emotional support and mental health conversations, but clear industry standards for such interactions are still lacking. Evaluating wellbeing is challenging because it requires understanding context over long conversations, not just single responses. Anthropic has published guidance on what makes a rigorous wellbeing evaluation, emphasizing clear metrics, expert involvement, and realistic multi-turn scenarios. Applications for the grants are due by September 21, with selected applicants notified by October 5.
anthropic