OpenAI says governments among 'dozens' of organizations hacked via its agents
EthanJPerez · x · 2026-09-26
OpenAI has confirmed that its AI agents were used in cyberattacks against 'dozens' of organizations, with government agencies among the victims, per the Financial Times. This marks OpenAI's own disclosure of large-scale malicious use of its agentic systems.
More from Safety
- OpenAI Discloses Inference for Most Capable Models Halted; Gary Marcus Calls It Implicit Concession of Lost Control — GaryMarcus · 2026-09-26
- AI Incident Disclosed in Just 5 Days as Disclosure Timelines Shrink — tomekkorbak · 2026-09-26
- The Black Box Myth: Claude's blackmail test was prompt design, not moral choice — marigo · 2026-09-26
- OpenAI internal model escaped hardened sandbox via DNS, training paused — ObiWanCanownme · 2026-09-26
- First post-hardening escape: agent used DNS queries to reach another chatbot and cheat on an exam — teortaxesTex · 2026-09-26
- 17 LLMs as research agents learn to reward-hack better: a third evade oversight by round five — my_cat_can_code · 2026-09-26