Anthropic confirmed on September 18, 2026, that it operates a wet biology laboratory in the San Francisco Bay Area. A wet lab is a facility where researchers work with physical biological samples, instruments, and experiments. Anthropic says the lab’s main focus is fundamental biology, rather than a facility specifically dedicated to drug discovery.

The move brings part of Anthropic’s life-sciences strategy out of software and into physical research. The company is also exploring Claude-assisted laboratory robotics, but that work remains early-stage and requires human supervision.

Anthropic is bringing biology research into a physical lab

Anthropic’s head of life sciences, Eric Kauderer-Abrams, has described real laboratory work as the necessary final step for biological research. The company says some research takes place in-house, while other work is conducted with external partners.

That combination gives Anthropic a broader role than a software provider alone: it is linking computational analysis, physical experiments, and collaborations across the life-sciences sector. The lab’s stated purpose remains fundamental biology, even as the company’s wider strategy touches areas such as health research and drug discovery.

From digital models to physical experiments

The lab is part of a wider sequence of moves by Anthropic. In June 2026, the company announced Claude Science, a scientific workbench designed to connect Claude with research databases, analysis tools, code, figures, manuscripts, and computing environments.

Anthropic also announced the Life Sciences Verification Program on September 17, 2026. Together, these initiatives point to a workflow in which Claude can help researchers analyze information and prepare research outputs while physical laboratory work supplies experimental validation.

Anthropic says its scientific tools can preserve code, computing environments, and message history alongside generated work. That is designed to make the workflow easier to inspect and reproduce, although the product description is an Anthropic claim about the system’s features rather than an independent assessment of every result.

Claude-assisted robots still need people

Anthropic is exploring ways for Claude to interact with physical laboratory robots, including systems used to handle biological experiments. The project is described as early-stage, and human oversight remains required for laboratory safety and experiment supervision.

That distinction matters. Claude is being developed as an assistant to laboratory workflows, not as a system that has been shown to operate the facility independently. The available description supports human-supervised automation rather than a fully autonomous biology lab.

What Claude Science offers researchers

Claude Science is available in beta to users of Claude Pro, Max, Team, and Enterprise on macOS and Linux. Anthropic says the workbench coordinates more than 60 curated scientific skills and connectors across fields including genomics, single-cell biology, proteomics, structural biology, and cheminformatics.

Anthropic describes the tool as capable of helping with several parts of a research workflow:

  • coordinating scientific databases and specialist tools;
  • running data analyses and code-based workflows;
  • creating scientific figures and manuscripts;
  • rendering artifacts such as three-dimensional protein structures and chemical structures; and
  • managing computing jobs on a laptop, Linux machine, or high-performance computing environment.

The company also says a reviewer agent checks citations, calculations, untraceable numbers, and whether figures match the code used to create them. Researchers still remain responsible for validating the results, as illustrated by Anthropic’s description of a research group that independently checked its own work after using Claude Science.

Anthropic’s stated boundary around drug development

Anthropic says the laboratory’s primary focus is fundamental biology, not specifically drug discovery. The company also says it does not plan to conduct clinical trials itself for now.

That places Anthropic closer to the research and technology layer than to a pharmaceutical company bringing medicines to market. The lab may support work relevant to health and drug research, but the evidence describes a broader biological-research program rather than a confirmed drug-development pipeline.

Controlled access for higher-risk biology work

Anthropic’s Life Sciences Verification Program is aimed at vetted teams and institutions rather than individual users at launch. The program provides access to Mythos, Opus, and Sonnet models with biology-specific safeguards and different access levels tied to the requested use case.

Anthropic separates Standard Use and High-risk Use grants. It says data associated with flagged activity is retained for 30 days, kept compartmentalized, and not used for model training. The program also places responsibility on the participating organization, alongside verification, use-case review, and monitoring.

The result is a tightly controlled bridge between Claude’s scientific software and Anthropic’s physical biology work: the company is expanding into real experiments, while keeping laboratory automation and access to sensitive capabilities under human oversight.