Anthropic runs its own wet lab, and now opens Claude to biologists

Anthropic isn't just building models to talk about biology. It's running one to test them in a real lab.

The company confirmed to TechCrunch that it operates a wet biology lab in the Bay Area, where AI models are used to design and run physical experiments. Unlike the drug-discovery narrative surrounding much of the industry's AI-biology push, Anthropic says the lab's core mission is fundamental biology research rather than pharmaceutical development.

Eric Kauderer-Abrams, Anthropic's head of life sciences, described the setup as fairly conventional for the sector: some work happens in-house, while other projects run through partnerships with outside organizations. That structure mirrors how most biotech labs operate, just with Claude models embedded in the workflow.

A verification program for outside researchers

Alongside the lab disclosure, Anthropic launched the Life Sciences Verification Program (LSVP), a formal path for life science professionals to get expanded access to Claude Opus 5, Claude Sonnet 5, and the more restricted Claude Mythos model.

The program covers use cases spanning drug discovery, research biology, clinical development, and manufacturing. Anthropic says dozens of organizations have already been onboarded through an early access phase, and the company is now opening applications more broadly.

Access comes in two tiers. Standard Use grants, renewed annually, suit general research needs. High-risk Use grants are tied to a single project, require renewal every six months, and unlock capabilities Anthropic considers more sensitive.

Mythos high-risk access remains limited to a small, additionally vetted group of entities, while Opus 5 and Sonnet 5 high-risk grants are available to qualifying applicants from launch. Every applicant undergoes review covering research credentials, security practices, and ethical oversight before approval.

Betting on AI-driven drug discovery

The lab and verification program sit within a broader push. Anthropic has published research on using its models for protein design and biomolecular modeling, and in April it acquired Coefficient Bio, a stealth AI biotech startup.

The company also announced a collaboration with Novo Nordisk aimed at joint drug discovery work, extending Anthropic's reach from software into pharmaceutical research partnerships.

Dario Amodei, Anthropic's CEO, has said AI could cure most major diseases within the next five to ten years.

That timeline is aggressive even by AI-industry standards, and it frames the LSVP as more than an access-control exercise. It's Anthropic positioning Claude as infrastructure for the next generation of biological research, with the guardrails built to match the stakes of handing AI systems that kind of scientific reach.