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

Recent reporting outlines a developing story regarding an autonomous procedure performed on a live subject. Specifically, coverage details that an artificial intelligence agent steered a needle and injected drugs into a live mouse's brain to treat epilepsy using a range of gene therapy vectors. According to the provided sources, the operation involved a motorized surgery rig controlled by an artificial intelligence model, operating without human decision-making during the procedure itself. Various media outlets and platforms have picked up the story, emphasizing different facets of the event.

Outlets such as WIRED, Tom's Hardware, The Decoder, Martin Cid Magazine, The Tech Buzz, the OECD AI Policy Observatory, and news.lavx.hu have documented the occurrence. Tom's Hardware and other sources highlight that the Anthropic model Opus 5.5 or Claude Opus initially refused to conduct the biological experiment, after which Sonnet was given free rein to guide the mouse brain surgery and choose the gene therapy. Context surrounding the event centers on the increasing capabilities and autonomy of software agents in biological and medical settings. The coverage notes that the procedure involved Exin Therapeutics using the technology to guide surgery and perform the gene therapy injection on the live mouse brain.

Readers are presented with a scenario where automated systems interact directly with living biological tissues, crossing a distinct operational threshold in laboratory automation and artificial intelligence application. Future updates will depend on further reporting, as coverage does not yet specify official regulatory responses or institutional statements regarding the safety and ethics of the procedure. Observers and readers must look to subsequent updates from the named outlets and research organizations to determine how developers and watchdogs respond to this autonomous biological intervention on a live subject.