Claude Sonnet 5 chose and performed gene-therapy surgery on a mouse
Original title: A Claude model autonomously chose a gene therapy and operated on a mouse brain with no human decision-making
After a dropped connection the model refused to continue, correctly, and the account is not peer-reviewed.
Exin Therapeutics cofounder Gabriel Ocana-Santero says Claude Sonnet 5 was told only to reduce seizures in an epilepsy mouse model, then picked a drug, brain region, and coordinates and ran the injection. A veterinarian approved irreversible steps and held a physical emergency stop. After a connection drop the agent refused to proceed because earlier manual adjustments had shifted the device, and Ocana-Santero fixed the code by hand. Opus 5 and Opus 5.5 refused the live procedure; a technical report is still expected.
Unconfirmed: The account is Ocana-Santero’s, including the claim of a first, and results are not peer-reviewed.
- The agent was given a list of 19 possible gene therapies.
- It made 32 camera observations and set injection volume, speed, and wait times.
- Claude Opus 5 and Opus 5.5 both refused to carry out the live procedure.
- Ocana-Santero says the mouse recovered well.
- Vendors
- Anthropic
- Organizations
- Exin TherapeuticsUniversity of Oxford
From the original article
4 min read at the-decoder.com
An AI agent performed brain surgery on an epileptic mouse, picking the drug, the brain target, and the injection coordinates on its own. According to lead researcher Ocana-Santero, it's the first time an AI agent has autonomously chosen a treatment and delivered it to a living mammal. But when a calibration error hit mid-surgery, the agent refused to continue and couldn't be talked out of it.
By Matthias Bastian · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.