A debate over agentic AI safety asks whether the field can avoid repeating past technology risks

JJitsev · x · 2026-07-27

@JJitsev argues that major technologies have historically prompted parallel safety research — like nuclear, bio, or chemistry — and asks why agentic AI should be different if the stakes are so high.

@BlancheMinerva replies that they have not seen evidence anyone is willing to put in the necessary security precautions to make sufficiently powerful agentic AI research safe, and says that is why they are increasingly convinced nobody will.

Related event: AI Ethics Debate: Is Training Powerful Agentic AI Immoral(7 posts)→

Original post →

More from AGI Musings

AGI Musings channel →