Anthropic opens AI biology lab as Claude guides experiments

Anthropic moving from pure software into a staffed biology wet lab is a big milestone in the race toward more capable, agentic AI. Up to now, most frontier work in AI for science has stopped at simulations and literature analysis. By wiring Claude directly into tools and workflows that run physical experiments, Anthropic is closing the loop between model predictions and real world biological feedback. That will almost certainly accelerate drug discovery and basic biology, but it also creates a live testbed for AI systems that plan, adapt and iterate in the physical world.

Strategically, this puts Anthropic on a different trajectory from labs that only sell APIs or generic copilots. The company is building deep vertical expertise in life sciences, helped by its Coefficient Bio acquisition and partnerships like Novo Nordisk, while also launching a Life Sciences Verification Program to court external researchers. That combination of proprietary wet lab infrastructure, domain specific talent and frontier models could make Anthropic the reference platform for AI driven biology.

The uncomfortable part is safety. This lab appears just as Anthropic leaders talk openly about greater than 10 percent extinction risk from AI and call for slowing development. Giving increasingly agentic models pathways into real experiments raises biosecurity stakes and will intensify calls for hardware standards, access controls and third party oversight in AI biology. In other words, this move pushes both capabilities and regulation pressure forward at the same time.