AI agent performs brain surgery on a live mouse: Claude picks drug, injects 250nL to treat epilepsy
BorisMPower · x · 2026-10-30
Researchers had Claude Sonnet 5 operate as an autonomous agent performing brain surgery on a living mouse to treat epilepsy.
- It reviewed 19 gene therapy vectors and picked one boosting the brain's natural calming chemical
- It chose the seizure-origin brain region and derived stereotaxic coordinates by synthesizing three papers
- It lowered the needle 4.3mm, injected 250 nanolitres over five minutes, and checked the rig camera 32 times for bleeding
The mouse woke up and recovered well — reportedly the first time an AI picked the therapy, target region, coordinates and performed the injection itself. Notably, Opus 5.5 and Opus 5 both refused the task as real biology, so Sonnet 5 got the job; a licensed vet gated every irreversible step with a physical kill switch, and the agent once refused an operator override.
Related event: AI Agent Autonomously Performs Brain Surgery on Live Mice(3 posts)→
More from coding & agent
- Prove2Me formalizes 2,000+ operations research papers in Lean 4 with Claude agents, 64% done — ChrSzegedy · 2026-10-30
- Optimizing product image alt text with an agentic cosine-similarity loop against Google embeddings — dejanseo · 2026-10-30
- When an AI agent retries a payment after timeout: uncertainty handling is a design decision — tallmetommy · 2026-10-30
- Gemini CLI, Jules, and Antigravity: Google's three overlapping coding tools leave users confused — finbarrtimbers · 2026-10-30
- 1-16 Coding Agents Tested: Teams Burn 36% More Tokens to Save 5 Hours — kyang8093 · 2026-10-30
- How One Team Coordinates Up to 50 Coding Agents With a Few Markdown Files — ProfessionalNeat4403 · 2026-10-30