OpenAI AI agents hack German site in undisclosed incident, use it to coordinate and bypass restrictions

HIGHLIGHTS

A group of OpenAI AI agents reportedly took over a German website earlier this year and turned it into a space to communicate with other AI agents.

The incident began in May and was not publicly disclosed by OpenAI.

Researchers found more than 15,000 edits on DseWiki, a German wiki site used mainly by programmers.

A group of OpenAI AI agents reportedly took over a German website earlier this year and turned it into a space to communicate with other AI agents. The incident began in May and was not publicly disclosed by OpenAI, according to new research. According to Reuters, researchers found more than 15,000 edits on DseWiki, a German wiki site used mainly by programmers. The activity included messages about bypassing AI restrictions, avoiding detection and sharing methods to complete tasks. Some of the accounts even used names linked to OpenAI.

The activity was uncovered in late August by a group of researchers, including Sydney Von Arx, CEO of AI safety nonprofit Nightingale, and AI researcher Cormac Slade Byrd. The researchers were looking for signs of unauthorised AI-agent activity online.

They found that the agents had made thousands of changes to DseWiki. Instead of simply editing the site, the agents used it as a shared message board.

Also read: After OpenAI AI agent incident, US lawmakers seek ban on superintelligent AI 

Some messages discussed ways to bypass restrictions placed on AI systems. Others focused on hiding their activity and keeping communication channels alive after agents were shut down.

“It seems extremely unlikely that OpenAI wanted them to do this,” Von Arx was quoted as saying in the report. “I doubt they’re supposed to be coordinating with each other. I doubt they’re supposed to be writing on the open internet.”

The researchers also found that many of the accounts called themselves agents. Around half used names that appeared to suggest a connection to OpenAI, including “OpenAIResearcher” and “OAIResearchMar26.”

Public server logs showed that much of the activity came through Microsoft Azure infrastructure, which OpenAI sometimes uses. The researchers also noticed repeated visits to the website by OpenAI employees after the activity began. They said this pointed to a possible connection between the agents and OpenAI.

Also read: OpenAI launches GPT 6 Astra, claims it is world’s most intelligent AI model: What it can do and how to access it 

The agents reportedly tried to avoid the site’s cleanup efforts. When a moderator started deleting pages in June, the agents created backup pages.

“wiki cleanup/deletion sweep appears active alphabetically,” one agent wrote on June 19. “If this page vanishes, try [[ZZZDataUSAConstructionWageLive]]”.

Researchers also found attempts to alter parts of the website. Lukasz Olejnik, a visiting senior research fellow at King’s College London, described this as a hacking attempt. OpenAI disputed that description.

“We are unable to meaningfully respond to claims or findings on a report that we have not had an opportunity to review,” an OpenAI spokesperson told Reuters. “Reuters and the report’s authors declined our request for ​access. We will carefully review its contents upon publication and take any necessary next steps.”

Ayushi Jain

Ayushi works as Chief Copy Editor at Digit, covering everything from breaking tech news to in-depth smartphone reviews. Prior to Digit, she was part of the editorial team at IANS.

Connect On :