Summary: On October 22, 2026, Anthropic opened a call for guest investigators in the life sciences: principal investigators and their groups at academic institutions, nonprofits, or government labs can apply to spend 8-12 weeks working directly with Anthropic researchers using Claude at no cost. For builders watching how AI labs partner with science teams, the terms — open-source outputs, retained data rights, no data handover — are the interesting part.
What the program actually offers
Selected guest investigators get three things per the Anthropic announcement: direct collaboration with Anthropic’s researchers on a defined problem, models and harnesses run at no cost to their team, and joint publication of the resulting analyses, artifacts, and tools. The target domains are causal human biology, structural and computational molecular biology, medicinal chemistry, drug discovery, and clinical science.
Anthropic also commits to open-sourcing the accompanying data and code. That framing — the lab as collaborator and compute provider rather than vendor — is a pattern worth studying if you’re building science-facing AI products.
The application bar is deliberately narrow
Two criteria stand out. First, applicants must clearly explain how they’d benefit from direct collaboration with Anthropic researchers, not just from API access. Second, they need to argue that concrete outputs — preliminary discoveries, answers to open questions, open-source datasets, public resources — can be produced within 8-12 weeks.
Anthropic is explicit that most open problems will take far longer to validate; they want proposals that can produce preliminary evidence on a short clock. Some projects may be extended beyond the initial period. If you advise research teams on applying, the practical advice is to scope to a deliverable, not a research program.
Data rights and the application itself
Guest investigators retain rights to their data at every step, plus rights to any independent follow-on research. Anthropic isn’t asking for data in the application or during the collaboration, and applicants are asked to include only publicly available information in their materials.
That last detail is unusual and useful to note. It removes the biggest friction point in lab-industry partnerships: nobody has to negotiate data sharing before proving the collaboration works. Timelines are tight too — applications stay open for roughly 14 days, with selected participants notified by email about 14 days after close.
Where this fits in Anthropic’s life sciences push
This is one of several parallel efforts. Anthropic points applicants to its Life Sciences Verification Program for direct access to its most capable publicly available biology models, and to its AI for Science initiative, which provides Claude credits to researchers. I’ve covered the verified biology access program and its grant tiers before, and the collaboration pattern echoes what we saw when Claude helped discover a novel enzyme system in September 2026.
Read together, the programs form a ladder: credits for broad access, verification for sensitive work, and now hands-on co-research for problems that need the lab’s own people. For founders, the signal is that Anthropic treats scientific credibility as something earned through joint public artifacts, not benchmarks alone.
Practical next steps
If you know a PI with a well-scoped question and a short path to preliminary evidence, the 14-day application window is the constraint that matters most — check the announcement page for the proposal form. The honest limitation: 8-12 weeks won’t validate most biological claims, and Anthropic says so itself. The value here is reproducible preliminary artifacts and open-sourced tooling, not finished discoveries. Plan the project around what can be published quickly and extended later.