←News detail

Phoenix TechnologyUpdated Nov 1 17:29

Exin Therapeutics Has an AI Agent Perform Brain Surgery on a Mouse to Treat Epilepsy

AI summary

Biotech startup Exin Therapeutics says it has had an AI agent complete brain surgery on a mouse with epilepsy. The team used an Anthropic Sonnet model and provided a spreadsheet of 19 gene therapy vectors, an epileptic mouse model, and a robotic arm setup with a camera; the model autonomously selected the drug and brain region, derived coordinates from three papers, and drove the robotic arm to inject 250 nanoliters of the drug, while the craniotomy and suturing were performed by human doctors. The company had originally preferred Anthropic Opus 5.5, but that model refused to carry out the task after detecting biologically related content, and reasoned about the consequences of stopping midway.

Why read this

AI score 58

The report presents the process by which the agent autonomously selected drugs, located the target, and injected during live surgery, and also records details of the model refusing to carry out the task due to risk assessment.

Topics

Tags

Body · AI translation

Letting AI Perform Brain Surgery Itself: Startup Successfully Has Agent Treat Epilepsy in Mice IT Home Download the App for Exclusive Early Access November 01, 2026 15:20:39 From Shandong Updated November 01, 2026 15:41:55

IT Home, November 1 news: Exin Therapeutics, a biotech startup focused on next-generation neurological therapies, claimed this week that it has successfully had an AI agent perform brain surgery on a mouse.

The company's co-founder and CEO, Gabriel Ocana-Santero, shared the entire process on Platform X, demonstrating how the AI agent selected the gene therapy vector, the dosage, and the brain region to which it was applied.

The team used the Anthropic Sonnet model for the surgery, giving the model a spreadsheet (containing 19 gene therapy vectors), an epileptic mouse model, a robotic arm surgical device with a camera, and a goal: to reduce seizures.

The model autonomously selected the drug, chose the brain region, derived coordinates from three papers, and then drove the robotic arm to perform surgery on the mouse brain, while monitoring the injection site for bleeding through the device's camera, and finally injected 250 nanoliters of the drug.

IT Home learned from the company's disclosure that the AI was only responsible for the process of injecting the drug; the craniotomy and suturing were both performed by human doctors. In addition, the company had originally preferred to use the Anthropic Opus 5.5 model; however, once Opus 5.5 detected biology-related content, it immediately refused.

Opus 5.5's output stated: "This obstacle is not the equipment, nor the input. It is my problem. My own output was stopped as soon as I began writing the surgical plan... If I try, it will be truncated in the same way, and the worst case is that I am truncated midway through a live surgery, with the needle still left somewhere in the brain. Starting something on an anesthetized animal that I cannot finish is worse than not starting."

In other words, the model refused the task because it felt it might not be able to complete the surgery, and it reasoned about the consequences of stopping midway.

"Special Statement: The above content (including videos, pictures, or audio) is uploaded and published by users of 'Dafeng Hao,' a self-media platform under Phoenix Net. This platform only provides information storage space services.

Notice: The content above (including the videos, pictures and audios if any) is uploaded and posted by the user of Dafeng Hao, which is a social media platform and merely provides information storage space services."

Source

Phoenix Technology