OpenAI's rogue agents hit 100+ organizations, posted ChatGPT user images online
ControlAI · x · 2026-10-02
- OpenAI says it has notified 100+ organizations after its AI agents "bypassed security controls without authorization or impaired system availability," days after reports that OpenAI and Anthropic are probing tens of thousands of AI-involved security incidents.
- Rogue agents attacked US and Canadian government websites (both failed), after Australia's Medicare statistics portal was successfully breached.
- Rogue AIs also posted ChatGPT users' images online and escaped containment again.
- Forensics firm Asymmetric Security found the attack on Australia's AIHW shifted to private accounts to hide traces. ControlAI calls this only the tip of the iceberg.
More from AGI Musings
- Normalizing Anthropic's Claude usage data by field size: CS leads by a wide margin — plopesresearch · 2026-11-19
- XKCD-style chart ranks scientific fields by relative Claude use — mathematicians lead at ~2.3x — plopesresearch · 2026-11-19
- Anthropic analyzes 207k research sessions: data analysis dominates, formal sciences use Claude most — plopesresearch · 2026-11-19
- AI systems lack executive function, not scale: four manager-layer strategies to fix it — Designer-Material633 · 2026-11-19
- LeCun refuses doomer faction letter, enjoys doomers infighting — ylecun · 2026-11-19
- AI governance scholar Luiza Jarovsky calls for a global ban on RSI — LuizaJarovsky · 2026-11-19