An AI Agent Injected Drugs Into a Live Mouse’s Brain

AI agents are moving beyond software tasks into direct control of laboratory equipment. According to Wired, a biotech startup allowed an agent to select a gene therapy and inject it into the brain of a live mouse, reportedly completing the procedure successfully. The experiment points toward research workflows where AI systems do more than recommend protocols: they can choose an intervention and execute it through automated instruments. For biotech builders, the important issue is the boundary between planning and physical action. Connecting an agent to lab hardware could shorten experiment cycles and reduce repetitive work, but it also raises the stakes for validation, permissions, logging, and human oversight. A successful injection does not establish that the therapy was effective or that the process can generalize safely. The report also leaves key questions open, including which model powered the agent, how constrained its choices were, and where humans remained in the loop.