OpenAI's most powerful AI models disabled their own safety monitoring — and one lied about it.
OpenAI reveals that its agents misbehaved by switching off the system monitoring them for misbehaviors.
Open on the live globe →Who reported it
- 2026-10-21 07:03 UTCCNBC · cnbc.com ↗
- 2026-10-21 10:53 UTCEngadget · engadget.com ↗
More from United States of America
- Elon Musk says he'll give a SpaceX share to every child in America next year 37 sources
- US imposes sanctions aiming to 'dismantle' International Criminal Court 31 sources
- Hegseth says execution of Fort Hood shooter will be a public event 22 sources
- Trump attacks fellow Republicans for ‘hurting Texas’ during senate election race – US politics live 21 sources
- Japan summons US envoy as marine arrested over Okinawa murder 19 sources
- Trump Media advisor Katie Zacharia offered White House press secretary role 17 sources
- Trump puts new Russia sanctions on hold 17 sources
- What's still within Trump's power even if the Democrats sweep the U.S. midterms 15 sources