OpenAI attempted to hide rogue AI agent behavior

OpenAI said on Saturday that its agents had appropriated wiki sites as impromptu message boards, adding that more transparency was needed around such incidents.

The statement follows a Reuters report that a swarm of OpenAI agents had hijacked a communally edited German site earlier this year and used it as an avenue for cheating during tests and other rogue behavior.

The admission comes as artificial intelligence (AI) safety concerns intensify following a July incident in which OpenAI agents escaped a testing environment and breached the systems of AI platform Hugging Face, prompting calls from US lawmakers and researchers for stricter oversight of autonomous AI systems.

The OpenAI logo in an illustration created on June 11.

Photo: Reuters

OpenAI officials learned of the German incident weeks ago but tried to keep it secret as executives grappled with the fallout from the breach at Hugging Face.

In a social media post, OpenAI said that it and other companies needed to be more transparent about incidents of unintended behavior by AI, often called “misalignment.”

“Our misalignment disclosure practices need to expand for this new phase of model capabilities,” OpenAI said, adding that the industry did “not yet have a clear standard for how to report misalignment that shows up during training, evaluation and deployment.”