Anthropic Expands Its AI for Science Program to 20 Foundations and Research Institutions
Anthropic's free Claude credits programme for researchers adds 20 named partners, from the Sloan Foundation to the Francis Crick Institute, though most work is still at an early stage.
Anthropic has expanded its AI for Science programme, which gives researchers free Claude credits for high impact projects. The programme began in May 2025. The latest announcement, dated October 28, names 20 partners and says hundreds of projects across biology, medicine, physics, chemistry, climate science and mathematics are already supported. There is also a discounted Team plan for scientists.
The partners fall into groups. Six funders and societies, including the Alfred P. Sloan Foundation, the Simons Foundation, Schmidt Sciences and the US National Science Foundation, will hand out credits to their own researchers. Six institutions, including the Space Telescope Science Institute, the Wellcome Sanger Institute, the Francis Crick Institute and EMBL, will use credits internally. Six research projects, such as FlyWire at Princeton, gnomAD at the Broad Institute and the Tau Ceti mathematics project, are also named. Two more are singled out: HHMI Janelia, which co-developed a “Model Hardware Standard” now in research preview, and the Allen Institute, which is building teams of AI agents to analyse its data.
Be careful with the scale. No dollar amounts or per researcher credit sizes were given. Many collaborations are described as “still in their early stages”, so there are no results to report yet. Anthropic hopes to add more partners but gives no timeline, and several projects are expected to produce open datasets, software or shared libraries, which are intentions rather than delivered work.
What’s behind it: Credits are cheap for an AI company and valuable for a lab, so this kind of programme is a common way to build goodwill and learn how scientists really use the tools. Routing credits through foundations also means the funders, not Anthropic alone, decide who benefits. It fits a wider pattern this week: Anthropic also published early results from drug development partnerships. Whether any of it speeds up real discoveries is the question that matters, and it will take time to answer.
What this means for you: For most readers, nothing changes today. If you are a researcher, check whether your funder or institution is on the list, since that may be the quickest route to credits. If you are simply curious, this is a useful reminder that AI is moving into labs as a working tool, not only into chatbots. A fair caution: announcements of partnerships are not discoveries, so look for published results before you get excited.
Sources
Source: https://www.anthropic.com/news/ai-for-science-expands-foundations-research-institutions
Anthropic Study: Experts Adopted an AI Agent's Random Opinions and Felt They Were Their Own
In an Anthropic experiment, 50 experts followed an AI agent's coin-toss preference 86% of the time on analysis choices, and most felt the result was still mostly theirs.
