Anthropic runs a laboratory carrying out biological research experiments

Anthropic has confirmed it runs a wet biology laboratory in the San Francisco Bay Area, where its AI models are used to carry out physical experiments. The revelation connects the company's public promises about AI transforming medicine with hands-on scientific work — and raises questions given its own warnings about AI's catastrophic risks.

Testing Theories in the Real World

Dario Amodei, Anthropic's CEO, argued just last week that "AI could cure most major diseases in the next 5–10 years." For that vision to become reality, an AI system needs ways to validate its hypotheses in physical experiments rather than only in silico.

"We believe that to do biology, the final test is still, and will be for a while, in real lab work," Eric Kauderer-Abrams, Anthropic's head of life sciences, told Reuters. "We absolutely are doing that today." He explained that the facility functions like a typical biotech lab: Anthropic runs some of its own research there while also collaborating with outside partners.

The confirmation should not come as a complete surprise. In April, Anthropic acquired Coefficient Bio, a stealth-mode AI biotech company.

Fundamental Biology, Not Drug Discovery

Anthropic declined to detail the lab's projects but said its primary emphasis is fundamental biology rather than discovering new drugs. That distinction matters strategically: the company counts major pharmaceutical firms among its customers and partners, including Novo Nordisk, with which it recently announced a joint drug discovery agreement. Anthropic has previously drawn criticism for launching products seen as competing with its own customers' offerings.

Alongside the lab, the company this week introduced a Life Sciences Verification Program, granting vetted biosecurity researchers access to its most capable models. It has also released recent reports on work supporting drug discovery, including research on accelerating protein design and on improving biomolecular modeling.

Warnings Versus Experiments

The lab news lands amid heightened anxiety within the company itself. Anthropic researcher Jacob Coxon recently resigned, warning that "the people building AI earnestly believe that it could kill us all by the end of the decade." Anthropic's own alignment lead has put the odds of AI exterminating humanity within ten years at greater than 10%.

Amodei has called bioterrorism one of AI's most serious dangers, and last weekend published a post urging the industry to slow down and adopt self-regulation. The contrast between those dire statements and the new wet lab has not escaped observers. Investor and AI coding startup founder Chamath Palihapitiya wrote on X, only half-jokingly: "The group behind such hits as: 'We're All Going To Die' and 'Regulate Me Now' are building a wet lab in SF. I do not recommend this."

How Anthropic reconciles its safety advocacy with physical bio experiments — and how regulators respond — will be worth watching.

Meta description: Anthropic confirms a Bay Area wet biology lab where its AI runs physical experiments, even as its leaders warn of catastrophic AI risks.

Tags: Anthropic, wet lab, biotechnology, Dario Amodei, AI safety

Featured image: Abstract illustration of laboratory glassware intertwined with glowing neural-network patterns in cool blue and green tones, no people or logos.