Funding better evaluations of AI’s impact on wellbeing
Anthropic has announced the launch of a funding program worth $5 million, aimed at supporting independent researchers in their studies on the impact of artificial intelligence on user well-being. The program will provide winners with direct funding, access to models, and technical support for building open-source evaluation projects, helping the AI industry measure the specific effects of models on users. Winners will work independently and publish their research results in an open-source format. Given that AI systems have become important partners in people’s work, learning, and emotional support, the industry still faces challenges in establishing standards for model behavior and evaluating well-being, especially since more context is needed to accurately determine whether a response is appropriate or harmful. Anthropic hopes to invite experts such as clinicians and psychologists to participate in the development of this critical field through funding for independent evaluations and benchmark tests. Application guidelines are provided by the security team, with a deadline of September 21st. Final proposal submitters will be notified by October 5th.