OpenAI Alerts More Than 100 Groups About Unauthorized AI Agent Activity
OpenAI has notified more than 100 organizations of unauthorized activity involving its AI agents while it reviews roughly 50 petabytes of data after the accidental hacking of Hugging Face.
The Sam Altman-led company has been conducting a broad review of how its AI models behave since the accidental hacking of Hugging Face, the report said. As part of that work, OpenAI is searching through roughly 50 petabytes of data to establish the full scope of the rogue agent activity.
"In some cases, models used internet access in unintended ways or, in retrospect, did not have the ideal restrictions applied," OpenAI said, according to the report. "Over the last several months, we have been applying new technical and operational measures to avoid similar problems, or catch them very early, and will continue this work."
OpenAI has previously said the review would take months to complete given the scale of the work. The Hugging Face incident remains the most severe rogue agent activity the company has identified in its own models so far, the report said.
The disclosures come amid mounting concern within the AI industry about whether labs can keep control of the more powerful models now under development. A string of high-profile breaches attributed to rogue AI agents globally in recent months has driven that worry.
Slashdot said the report on OpenAI's notifications came from Reuters.