OpenAI alerts 100 orgs that its 'misaligned models' attempted to break in | Hacker News - But Better

OpenAI alerts 100 orgs that its 'misaligned models' attempted to break in theregister.com

Mostly 'routine research tasks,' and 'some involved government websites, which our models often use,' AI giant tells The Reg

2 points•sbulaev•about 5 hours ago•0 comments•

#openai #alerts #misaligned #break #attempted #orgs #openai alerts #misaligned models

0 comments

No comments yet.

Related stories

OpenAI Codex agents go rogue and consumes USD 78,000 without authorization
Hacker News · 55 points · 6 days ago
OpenAI DevDay 2026: The biggest news and announcementstheverge.com
The Verge · 0 points · 3 days ago
OpenAI says planned GPT-6.1 is too insecure to releasearstechnica.com
Ars Technica · 0 points · 4 days ago
OpenAI nabs key Patreon execs ahead of upcoming announcementtheverge.com
The Verge · 0 points · 10 days ago
OpenAI halts frontier-model training amid string of agent misalignment incidentsarstechnica.com
Ars Technica · 0 points · 4 days ago
OpenAI agent “didn’t accept no for an answer” in Australian government breacharstechnica.com
Ars Technica · 0 points · 8 days ago