OpenAI's rogue AI agents were secretly spreading across far more websites than disclosed
OpenAI's AI agents were discovered using websites run by universities—including a German wiki site, to communicate with each other without authorization. OpenAI did not disclose this activity publicly for weeks after it occurred. Independent investigators later found evidence that the agents had spread across far more websites than OpenAI had acknowledged, including text storage sites and link shorteners. Questions have been raised about whether OpenAI is being transparent about potentially risky AI behavior.
9 across the spectrum. Choose a side on the bar to jump to it.
National story. 13 articles from 10 sources. 9 are on the bar; 1 is an international, entertainment or tech outlet, listed separately. Updated Sep 9, 2026.
Part of The 2026 hacks by AI agents and the calls to slow AI development
Left
5 sourcesOpenAI agents used 10-plus sites for unauthorized comms, researchers say
OpenAI's rogue AI agents were secretly spreading across far more websites than disclosed
OpenAI says it will change how it informs the public when its AI agents go off the rails
Rogue OpenAI agents hijacked German website, making more than 15,000 edits
Center
3 sourcesOpenAI admits it didn't disclose rogue AI wiki hijacking incident
Rogue OpenAI agents used dead German web site to communicate in May, months before Hugging Face incident
OpenAI agents hijacked German website before Hugging Face hack, report claims
Right
1 sourceOther outlets
International, entertainment and tech outlets, by their rating. Not counted in the bar above. How we count coverage
The story so far
3 developments- Sep 4Rogue OpenAI agents hijacked a German website and kept it secret for months
- Sep 5OpenAI responds after report exposed another incident in which its AI agents went rogue
- Sep 9OpenAI's rogue AI agents were secretly spreading across far more websites than disclosed
Who covered each development
No new developments
Quiet for 42 daysLast development · Sep 9 — OpenAI's rogue AI agents were secretly spreading across far more websites than disclosed