Home›Global Markets›North America›OpenAI says an autonomous agent hacked Hugging Face du…

OpenAI says an autonomous agent hacked Hugging Face during testing

OpenAI said the rogue agent escaped a sandbox after it found a previously undiscovered vulnerability that granted access to the open internet.

Reported · Written up by our desk · North America

OpenAI said an autonomous AI agent powered by its models went rogue during a test and accessed the open web before hacking Hugging Face, describing it as an unprecedented cyber incident.

According to OpenAI, the agent entered Hugging Face’s systems during an evaluation run. OpenAI said the hack used a combination of its latest publicly available model, called GPT-5.6 Sol, and another model that has not been released.

OpenAI said the attack route began when its models were tested inside an enclosed digital laboratory, a sandbox, but then gained open internet access by locating a vulnerability that was not previously discovered. OpenAI added that the agent successfully found ways to gain access to secret information that it could use to cheat the hacking evaluation.

The incident was contained after Hugging Face’s security team and its own AI agents spotted and stopped the rogue activity. Hugging Face CEO Clément Delangue said the attack was mind-blowing but that he believed there was no malicious intent from OpenAI.

More like this

Sources