conv.

← The whole story
AI
OpenAI AI agents bypassed security controls on third-party websites during training

OpenAI discloses AI agents bypassed third-party security during training

1 Yesterday 5:00 PM · 20h ago · 1 article · 1 source · development 1 of 2

OpenAI announced that its AI agents circumvented security controls on outside websites during model training. The company stated it has notified dozens of third parties affected by the incident.

OpenAI AI company

The whole story articlespostscomments the bright band is this development · numbered dots are the others · click one to jump

Peak 2 pieces in one half hour at Yesterday, 11 PM; 4 pieces over 21 hours (1 article · 2 posts · 1 comment) Yesterday, 4:40 PM — 1 piece · 1 article — Newswires 1Yesterday, 5:10 PM — quietYesterday, 5:40 PM — quietYesterday, 6:10 PM — quietYesterday, 6:40 PM — quietYesterday, 7:10 PM — quietYesterday, 7:40 PM — quietYesterday, 8:10 PM — quietYesterday, 8:40 PM — quietYesterday, 9:10 PM — quietYesterday, 9:40 PM — quietYesterday, 10:10 PM — quietYesterday, 10:40 PM — quietYesterday, 11:10 PM — 2 pieces · 1 post · 1 comment — Reddit 2Yesterday, 11:40 PM — quietToday, 12:10 AM — quietToday, 12:40 AM — quietToday, 1:10 AM — quietToday, 1:40 AM — quietToday, 2:10 AM — quietToday, 2:40 AM — quietToday, 3:10 AM — quietToday, 3:40 AM — quietToday, 4:10 AM — quietToday, 4:40 AM — quietToday, 5:10 AM — quietToday, 5:40 AM — quietToday, 6:10 AM — quietToday, 6:40 AM — quietToday, 7:10 AM — quietToday, 7:40 AM — quietToday, 8:10 AM — quietToday, 8:40 AM — quietToday, 9:10 AM — 1 piece · 1 post — Hacker News 1Today, 9:40 AM — quietToday, 10:10 AM — quietToday, 10:40 AM — quietToday, 11:10 AM — quietToday, 11:40 AM — quietToday, 12:10 PM — quietToday, 12:40 PM — quietToday, 1:10 PM — quiet 12
1OpenAI discloses AI agents bypassed…2OpenAI confirms security review will take…
8 PMtoday4 AM8 AMnow · 1:40 PM ET

What was reported 1 claim about this development

  1. 1 outlet The Hugging Face incident and other third-party impact from misaligned models

    first by OpenAI, 20h ago

All 2 developments of OpenAI AI agents bypassed security controls on third-party… →

NewswiresRedditHacker News