Anthropic · 2026-08-25 · notable
Anthropic wellbeing grants — $5M for open evaluations of AI's effect on people
Anthropic is putting $5 million into outside research on how AI models affect user wellbeing. Grantees get funding, model access, and technical support, and publish their evaluations as open source. Applications close September 21.
Anthropic will fund outside teams to build open-source evaluations that measure how AI models affect the people using them.
What is it?
Anthropic's wellbeing research grants put $5 million behind independent evaluations of how AI models affect user wellbeing. Clinicians, psychologists, methodologists, and other subject-matter experts can apply. Applications close on September 21, and teams picked to send a full proposal hear back on October 5.
How does it work?
Selected teams receive direct funding, access to Anthropic's models, and technical support, and their work is published as open source so any developer can run it. A guidance document from Anthropic's Safeguards team, released with the announcement, sets out what qualifies: an evaluation should state plainly what it measures, involve clinical and subject-matter experts in its design and validation, and test for both precautions and harms.
Why does it matter?
Wellbeing is one of the hardest things to measure about a chat assistant, and public benchmarks mostly skip it in favor of reasoning, code, and factual recall. Paying for open-source wellbeing evaluations gives every lab, not only Anthropic, something concrete to test against. The guidance pushes in a useful direction too, asking for multi-turn conversations validated against real experts rather than one-shot prompts.
Who is it for?
clinicians, psychologists, and evaluation researchers
Try it
Applications close September 21 through the form linked from Anthropic's announcement.