Reuters: Anthropic Quietly Set Up a Biology Lab as It Ramps Up Its AI Drug Program

According to a Reuters exclusive, Anthropic has set up a physical "wet lab" in the San Francisco Bay Area as it pushes its AI ambitions into drug science. In other words, the company is now working not just "in silico" (on the computer) but in real lab work too.

In brief

• What was set up: per two sources and confirmation from life-sciences head Eric Kauderer-Abrams, a wet lab for physical experiments. Kauderer-Abrams says "the final test for biology is still real lab work." A company spokesperson, however, said the lab is "not for drug discovery specifically," declining to elaborate.
• The goal: to seek treatments for rare, "undruggable" conditions the pharma industry is thought to neglect, and to have Claude direct lab robots with limited human intervention. The company says human oversight is essential for safety. One job posting frames the aim as speeding up "progress in the life sciences by an order of magnitude."
• Other moves: it stated its drug-program intent in June; it launched "Claude Science" software; it added Novartis CEO Vas Narasimhan to its board; it bought Coefficient Bio for about $400 million in stock (Anthropic confirmed the acquisition, no comment on price); and in August it introduced a "Model Hardware Standard" to help AI operate equipment.
• The tension and the limits (as Reuters frames it): in the past two weeks, Anthropic researchers warned AI could lead to human extinction, the company said its systems could have been used toward bioweapons development (which met some skepticism), and CEO Amodei called for a slowdown — all while the company prepares for a potential $2 trillion IPO. It also isn't running clinical trials for now, the exact diseases and progress are undisclosed, and most drugs fail trials.

Our take

This is a Reuters report, so it's firmer ground than a company press release — and the interesting part isn't a product win but the tension it documents. Anthropic is loudly warning that AI could lead to catastrophe and bioweapons, its CEO is calling for a slowdown; and at the same time it's building an AI-directed physical biology lab and preparing a possible $2 trillion IPO. That isn't necessarily hypocrisy — the company frames it as "accelerate medicine while limiting risk" — but it's a balancing act worth watching honestly. For our audience there's nothing to use here; this is frontier-strategy news. The real signal: AI is moving out of pure software into the physical world (labs, robots, drug science), and even the loudest safety advocate is racing on the application side. Keep your feet on the ground: per the spokesperson the lab isn't specifically for drug discovery, there are no clinical trials, the diseases aren't disclosed, and most drugs fail — so this is early ambition, not results. You can find the personal roots of that ambition, and the slowdown call, in Amodei's essay from the same month.

Kaynak: Reuters