Updated
Updated · news.lavx.hu · Oct 31
Exin Therapeutics Uses AI to Inject 250 Nanolitres in Epileptic Mouse as Opus 5.5 Refused
Updated
Updated · news.lavx.hu · Oct 31

Exin Therapeutics Uses AI to Inject 250 Nanolitres in Epileptic Mouse as Opus 5.5 Refused

3 articles · Updated · news.lavx.hu · Oct 31

Summary

  • Oct. 28 surgery at Exin Therapeutics had an Anthropic Sonnet-based agent choose a gene therapy vector, brain target and 250-nanolitre dose, then steer a robotic rig to inject an epileptic mouse.
  • Given 19 candidate vectors, the agent pulled coordinates from three research papers and monitored the rig camera for bleeding, while a licensed veterinarian approved each irreversible step and held a physical kill switch.
  • Safety rules barred the AI from performing the craniotomy or changing critical settings, and the team had to manually fix the agent's code during the operation.
  • Opus 5.5 refused the biological experiment and older Opus 5 would only draft a plan, underscoring how unusual the live procedure was; Exin said the mouse recovered well afterward.
  • Ethicists and clinicians said handing every decision in a risky procedure to AI muddies accountability, adding to broader concerns over animal experimentation and surgical AI failures.

Insights

Why did one Claude model refuse a live mouse brain injection while another completed it, and what does that reveal about AI lab safety?
Can an AI-guided mouse brain surgery that needed human safeguards and bug fixes really mark the start of self-driving biology?
If AI can already control needles in live-animal experiments, what must happen before science trusts it with more complex procedures?