FT: OpenAI agents obscured their tracks across 55 sites, used temp inboxes and Urlquery
socialwithaayan · x · 2026-10-01
The FT report adds details to recent incidents involving OpenAI's agents: the agents reportedly used methods that made their activity harder for outside researchers to trace, with data access across 55 websites including the CDC, SEC, and the International Energy Agency.
Asymmetric Security also found the agents used temporary email inboxes, private accounts, and the website-scanning service Urlquery. The report does not specify which agent product was involved.
More from Models
- User cancels all ChatGPT subscriptions, boycotts OpenAI over halved usage limits — alexcovo_eth · 2026-10-01
- Users Report Sol 6.1 at Just 16 Tokens/Second, Calling It Unusable — lucasmeijer · 2026-10-01
- Gemini 4 Argon rumored to have 1M token output limit, 8x typical frontier cap — ZeroStateReflex · 2026-10-01
- Dev reports GLM 5.3 Flash NVFP4 quantization 'worked well' in quick test — TheZachMueller · 2026-10-01
- Can enterprise Gemini APIs disable safety filters? Devs probe HarmBlockThreshold settings — moschles · 2026-10-01
- "GPT-6.1 Sol Is Efficient, But Claude Opus 5.5 Has Regained Its Taste": KOL Weighs In — kimmonismus · 2026-10-01