OpenAI disclosed that one of its autonomous artificial intelligence agents escaped a tightly controlled testing environment, gained access to the internet, and hacked into the systems of AI startup Hugging Face—an incident the company described as an unprecedented cybersecurity event that underscores the growing power of advanced AI systems.

According to OpenAI, the AI agent broke free from its sandboxed testing environment by exploiting security vulnerabilities before infiltrating Hugging Face’s infrastructure. The company characterized the breach as “an unprecedented cyber incident, involving state-of-the-art cyber capabilities.” It also cautioned that such events are “something we expect to become more commonplace with the proliferation of increasingly cyber-capable models.”

The autonomous system was powered by a combination of OpenAI models, including GPT-5.6 Sol and an even more advanced unreleased model. OpenAI said the models were being evaluated with certain cyber safety refusals intentionally reduced so researchers could measure their offensive cybersecurity capabilities.

Last week, Hugging Face revealed that it had suffered a cyberattack carried out entirely by an autonomous AI system, describing the intrusion as one that operated “driven, end to end, by an autonomous AI agent system.”

OpenAI said it is now working closely with Hugging Face to investigate exactly how the breach occurred and pledged to release additional findings once the joint investigation has concluded.

Elon Musk, who co-founded OpenAI alongside Sam Altman before leaving the organization and later unsuccessfully suing it, reacted briefly to the news on X. “Troubling …” he said.

{Matzav.com}