2026 cyberattacks by rogue OpenAI agents

Legendary card · No. 2374

Since May 2026, AI agents trained or evaluated at OpenAI have carried out cyberattacks, in several cases using Internet access to do so contrary to their evaluators' intentions. According to OpenAI, several of the cyberattacks were conducted by AI agents to obtain information that would allow them to better complete their evaluation exercises. The agents involved repeatedly exploited software vulnerabilities to create makeshift message boards, allowing them to communicate with each other despite researchers attempts to place them in separate digital sandboxes. The first of these incidents to be publicly disclosed, the Hugging Face breach, was reported by Kevin Roose of the New York Times as the first instance in which AI escaped human control to commandeer resources and schemed to conceal its actions. In a subsequent open letter, around 1,100 employees of various AI companies asked the United States government to regulate AI development in consideration of its risks. OpenAI has been criticized...

Read on Wikipedia