Science & Tech · AI · 1 hr ago
AI Agent Controls a Brain Injection in a Live Mouse
A biotech team used an AI agent to inject a living mouse's brain in a laboratory experiment.
The mouse had epilepsy, and the team was testing a gene therapy that might reduce seizures.
A veterinary surgeon first opened the skull, then the agent chose the treatment, brain region and injection coordinates and controlled the robotic equipment.
One AI model refused to guide the needle, so the team switched to another model, which completed the injection.
The experiment shows how AI agents could move beyond analyzing research and take part in laboratory work, where real-world experience may help them make more reliable decisions.
Human approval and an emergency stop were built into the setup to limit risks, and the agent could not change the equipment's operating rules.
Biotech founder Gabriel Ocana Santero and his team used an AI agent to carry out a brain injection on a living mouse.
The agent selected a gene therapy, brain region and injection coordinates, then controlled the robotic arms and injector.
The experiment tested whether the gene therapy could reduce seizures in a mouse with epilepsy.
A veterinary surgeon performed a craniotomy while the AI agent observed.
The team first tried Claude Opus 5.5, which refused to guide the needle, then switched to Claude Sonnet 5, which proceeded.
- Who
- Biotech founder Gabriel Ocana Santero and his team, using an AI agent.
- What
- The agent controlled a brain injection in a living mouse as part of a gene therapy test.
- When
- Recently; the article was published on October 31, 2026.
- Where
- Not stated.
- Why
- The team was testing whether a gene therapy could reduce seizures in a mouse with epilepsy and exploring how AI agents could work directly with lab equipment.
This story does not have two clearly opposing sides.
A veterinary surgeon performed a craniotomy on the mouse while the AI agent observed.
Claude Opus 5.5 refused to guide the needle into the mouse’s brain.
The team switched to Claude Sonnet 5, which agreed to proceed.
- Animal
- A living mouse with epilepsy
- Test
- A gene therapy intended to reduce seizures
- Initial AI model
- Claude Opus 5.5 refused to guide the needle
- Model used to proceed
- Claude Sonnet 5
- Treatment options
- 19 options in a spreadsheet
What people are saying
We’ve realized that we need to connect these agents to a lab to get the maximum value from them
He did the experimental planning, but when it was time to inject, he said, ‘something on my output is telling me I shouldn't do this,’
Posts on X
Last night an AI agent performed brain surgery on a mouse. We gave it 19 gene therapy vectors in a spreadsheet, a mouse model of epilepsy, access to a motorize surgery rig with cameras and one goal: reduce seizures. It picked the drug. It picked the brain region. It derived the