Anthropic builds a San Francisco biology lab and puts Claude on the robots

Anthropic has built an in-house biology laboratory in the San Francisco Bay Area, and the company's head of biology, Eric Kauderer-Abrams, confirmed the facility's existence in an interview with Reuters on September 19. The ultimate test in conducting biological research is actual laboratory work, he said.

He framed the move around the company's mission of developing powerful AI to benefit the world, with the biggest opportunities seen in the life sciences and driving everything the company does. Anthropic, however, says it will not compete directly with traditional pharmaceutical and biotech companies.

Instead the work focuses on the stages before clinical trials. The initial emphasis is on discovering treatments for neglected diseases that drugmakers have largely overlooked, a space where AI-driven discovery can add value without colliding with commercial pipelines.

The lab is also being wired to the company's models. Anthropic is building a system that lets Claude directly control robotic equipment inside the laboratory, turning the model from an analysis tool into an operator of physical experiments.

Kauderer-Abrams said the company is mindful of the risk that AI models could be misused to develop biological weapons, and that safety measures will be deployed in parallel. We are balancing improving models to accelerate drug discovery with deploying them carefully to mitigate risks, he explained.

The laboratory follows a fast build-out of Anthropic's life-sciences ecosystem. The company has introduced Claude for Science, a tool aimed at scientists, acquired the biotech firm Coefficient Bio, and appointed Novartis CEO Vas Narasimhan to its board. In July it recruited John Jumper, a former senior staff research scientist at Google DeepMind who co-developed the protein-structure prediction AI AlphaFold and shared the Nobel Prize in Chemistry.

CEO Dario Amodei has also spoken publicly about losing his father to illness several years ago, which has been read as personal motivation behind the life-sciences push.

The bigger signal is that an AI lab is now running its own wet lab and wiring its model directly into robotic equipment, moving AI-for-science from papers to owned facilities. What to watch next is how specific Anthropic gets about research targets and biosecurity safeguards, and how much it publishes about Claude operating laboratory hardware.