Anthropic has opened a call for life scientists to spend 8 to 12 weeks working directly with its researchers. Principal investigators and their groups at universities, non-profits and government labs can propose projects in causal human biology, structural and computational molecular biology, medicinal chemistry, drug discovery or clinical science, and Anthropic will run Claude and its agent harnesses for them at no cost.
The terms are more generous than most industry tie-ups. Outputs are to be published jointly, with the accompanying data and code released as open source. Investigators keep the rights to their data throughout and to any independent follow-on research. Anthropic says it is not asking for anyone's data, either in the application or during the work, and asks applicants to include only public information in their proposals. Some projects may be extended beyond the initial period.
The catch is the timetable. Applicants must make a reasonable case that they can produce something useful within the window, such as preliminary findings, answers to open questions, datasets or public resources. Anthropic admits that most open problems take far longer to validate, and says it wants early evidence rather than finished results. Applications are open for about 14 days, with decisions promised roughly 14 days after they close.
The programme adds to a growing set of lab offers aimed at scientists. Anthropic already runs a Life Sciences Verification Program, which gives vetted researchers access to its most capable public models for biology, and an AI for Science scheme that hands out Claude credits. OpenAI, for its part, recently began offering academic labs up to 30 free seats with its GPT-Rosalind model for up to 12 months. Both companies want showcase results in biology, and both are paying for them in compute.
For a small lab without a compute budget, this is a low-risk way to see what Claude can do on a real problem with expert help on hand. The short window does shape what gets chosen, though. Projects that can show a quick, publishable signal will be favoured over slow, careful validation, and the joint publications will carry Anthropic's name as well as the lab's. Anthropic has not said how many projects it will accept or how it will judge them, so applicants should read the terms closely before committing their group's time.