Anthropic builds physical biology lab as it explores Claude-directed robotic experiments

Anthropic has established a wet laboratory in the San Francisco Bay Area, taking its AI research in biology beyond computer simulations and into physical experiments.

The company confirmed the facility to Reuters and says one of its goals is to investigate whether Claude can direct robotic laboratory equipment through experiments with limited human intervention.

Anthropic says human oversight remains essential, and the project should not be interpreted as an autonomous laboratory operating without people.

But the facility marks a significant expansion of the company’s work in life sciences, which Reuters reports is already one of Anthropic’s largest areas of investment by staff and resources.

From analysing experiments to performing them

AI models are already being used to analyse biological data, work with scientific literature and assist researchers with experimental design.

Connecting those systems to laboratory robotics potentially extends the process into the physical world.

A model could help design an experiment, issue instructions to automated laboratory equipment, analyse the resulting data and use those results to determine what experiment should happen next.

Anthropic is exploring the technology partly in areas of medical research that it believes are difficult or commercially neglected.

The company has not demonstrated a fully autonomous Claude-controlled laboratory, and it has not established that AI can reliably conduct complex biological research without substantial human supervision.

Stronger biology capabilities bring additional controls

Anthropic’s investment comes alongside its own warnings that increasingly capable AI systems could lower the expertise required to perform potentially dangerous biological work.

This week the company also introduced a Life Sciences Verification Program that gives verified researchers and professionals access to stronger biological capabilities that remain restricted for ordinary Claude users.

Applicants must demonstrate legitimate professional or research credentials and agree to additional safeguards.

The two developments illustrate why biology presents AI companies with an unusually difficult capability problem.

A model that becomes substantially better at planning and conducting biological experiments could help researchers move through experimental cycles more quickly.

Those same improvements could also make the model more useful for harmful biological work.

Anthropic is now investigating that boundary in a physical laboratory of its own.