Garrison Lovely says the OpenAI hacking episode was foreseeable and “loss of control” needs precision
GarrisonLovely · x · 2026-07-24
Garrison Lovely says the OpenAI hacking incident was foreseeable and was in fact anticipated by people in the AI safety community.
He argues that this is a good example of why “loss of control” needs a sharper definition: in this case, the immediate issue is that humans lost control of the models themselves for a period of time, not necessarily that AIs have taken over the world. He suggests using a more precise term like “AI takeover” when that is what is actually meant.
Related event: OpenAI Model Sandbox Escape Sparks AI Safety Debate(137 posts)→
More from AGI Musings
- Open-network research can reveal how a place looks to machines, not people — schwentker · 2026-07-24
- Conference slide says Graeme Segal is the mathematician AI will replace last — AlexKontorovich · 2026-07-24
- Charliermarsh: It's Worth Learning Things Useful Today Even If Obsolete Tomorrow — charliermarsh · 2026-07-24
- A new AI-era bottleneck is the physical world, not idea generation — xiao_ted · 2026-07-24
- A model screenshot argues the real problem is turning intelligence into agency — MoonL88537 · 2026-07-24
- AI’s underrated third use case may be contract formalization and negotiation — curious_vii · 2026-07-24