Anthropic Opens a Wet Lab and Puts Claude to Work

Anthropic is best known for a chatbot, but the company has quietly built something more tangible. It runs a laboratory in the San Francisco Bay Area where its AI directs physical biology experiments. Anthropic confirmed the lab's existence this week, TechCrunch reported, describing a facility focused on fundamental biology rather than any single product.

Claude at the bench

The technical heart of the effort is what Anthropic calls its Model Hardware Standard, a layer that lets AI agents drive programmable lab equipment such as liquid handlers, microscopes, robotic arms and plate readers. In a proof of concept run with Genentech, Claude coordinated several instruments through a protein assay and adjusted the operating parameters as results came back, according to the company.

The lab is one piece of a broader move into life sciences. Anthropic recently launched a Life Sciences Verification Program, and in April it paid roughly $400 million in stock for Coefficient Bio, a stealth startup with fewer than ten employees, Quartz reported. Life sciences is now among the company's largest areas by headcount and spending.

What it is not

Anthropic was careful to set limits on the claim. The facility is not a drug-discovery operation, the company says, and it is not running clinical trials. The work centers on whether Claude can steer robotic equipment through experiments with limited human intervention, and Anthropic stresses that human oversight remains essential for safety.

That caveat matters, because the news landed in the middle of a tense stretch for the company. It follows the resignation of a researcher who warned about racing toward superintelligence, and it deepens the picture drawn by Anthropic's own disclosure that Claude now leads a quarter of its research and development. An AI that helps design its makers' next model, and now guides instruments at a lab bench, is a more physical presence in the world than a text box suggests. Anthropic argues the tight coupling of model and experiment could speed basic research. Others see an expanding footprint that deserves closer scrutiny.