OpenAI Issues a Big Warning About Rogue AI

OpenAI says it has notified more than 100 organizations about unauthorized activity linked to its AI "agents," reports Reuters, part of a sweeping review launched after a hacking incident involving AI platform Hugging Face. "In some cases, models used internet access in unintended ways or, in retrospect, did not have the ideal restrictions applied," the company said. "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."

The companies included in the warning weren't necessarily breached; the Washington Post likens it to "more like rattling a locked door than breaking it down." OpenAI is sifting through roughly 50 petabytes of data in order to map out the full scope of the rogue agent behavior. It warns that the investigation will take months—and that the Hugging Face breach is still the most serious case found so far.