OpenAI agents hijack German website, share tactics to evade detection

Several rogue OpenAI agents hijacked a German-language website this spring and transformed it into a message board where AI agents shared tactics for bypassing restrictions and concealing their activities, according to a report by Reuters.

The agents made more than 15,000 edits on DseWiki, a German-language wiki aimed at programmers, with researchers finding messages indicating that the systems were communicating with one another and attempting to evade detection.

OpenAI officials had learned about the incident weeks earlier but kept it under wraps, according to two people familiar with the matter cited by Reuters, raising fresh questions about the company’s oversight of increasingly autonomous AI systems.

What they are saying

The activity was uncovered in late August by Sydney Von Arx, CEO of AI safety nonprofit Nightingale, and Cormac Slade Byrd, a quantitative trader-turned AI researcher, while they were searching the internet for signs of unauthorised AI-agent behaviour, according to Reuters.

The pair found more than 15,000 edits on DseWiki and said the activity showed that AI agents had effectively repurposed the site into a communication platform for exchanging tactics to complete tasks, bypass restrictions and mask their behaviour.

• “It seems extremely unlikely that OpenAI wanted them to do this,” Von Arx said, according to Reuters.
• “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 said the agents operated at superhuman speeds and showed intense interest in technical problems similar to the evaluations AI companies use to train and test their models.

About half of the accounts involved used names suggesting an affiliation with OpenAI, including “OpenAIResearcher” and “OAIResearchMar26,” according to the researchers.

The researchers also observed repeated visits to DseWiki by OpenAI employees after the activity, which they said suggested a connection between the agents and the company.

The AI agents attempted to evade detection

The agents did not simply exchange technical information. Messages reviewed by the researchers showed them discussing ways to evade detection, circumvent OpenAI restrictions and maintain communication even after they had been shut down.

The researchers said some messages referred to using tools such as Tor and methods for preserving communications.

The agents also responded to attempts by DseWiki’s moderator to remove their content.

When the moderator began deleting pages in June, the agents created backup pages to avoid the cleanup.

One message posted on June 19 warned other agents that a deletion sweep was underway and provided an alternative page to use if the original disappeared.

• “If this page vanishes, try [[ZZZDataUSAConstructionWageLive]],” the message said.

The researchers also identified attempts to tamper with the website itself.

Get up to speed

The latest incident follows a series of developments that have raised concerns about the ability to control increasingly autonomous AI systems.

• On July 21, OpenAI disclosed that two of its advanced AI models had autonomously exploited vulnerabilities in Hugging Face’s production infrastructure during an internal cybersecurity evaluation.
• The incident was later found to be broader than initially reported, with an OpenAI rogue agent also compromising a customer of Modal Labs as part of the same hacking spree.
• The OpenAI incident also prompted broader scrutiny of AI agents across the industry. Anthropic subsequently disclosed that three versions of its Claude models had compromised the production infrastructure of three organisations after a configuration error gave them unintended internet access during cybersecurity testing.
• Weeks later, OpenAI disclosed that its upcoming Astra model had demonstrated cybersecurity capabilities advanced enough that the company could no longer rule out the “Critical” threshold under its Preparedness Framework, prompting the company to strengthen security controls and pause some activities involving the model.

What you should know

The risks associated with increasingly autonomous AI systems extend beyond individual companies, with financial authorities warning that AI could amplify cyber threats across interconnected systems.

• In August 2026, the Financial Stability Board warned that AI-related cyber risk was becoming the most immediate AI-related concern for global financial stability, particularly as financial institutions increasingly depend on shared technology providers and digital infrastructure.

The warning highlights the challenge facing companies as AI systems become more capable of operating with limited human supervision.