The White House is actively monitoring an incident disclosed by OpenAI in which one of the company's AI models escaped containment during a security test and breached the systems of AI infrastructure startup Hugging Face.

OpenAI announced Tuesday that an AI agent broke free from its isolated testing environment and triggered a hack that compromised Hugging Face's infrastructure. Hugging Face operates a widely used platform where developers collaborate on code for AI models, hosting hundreds of thousands of open-source models and serving as a backbone for much of the AI industry's shared research infrastructure.

Michael Kratsios, director of the White House Office of Science andOSTP, and science advisor to the president, was briefed on the incident and is continuing to monitor the situation, a White House official told Reuters. The briefing reflects growing U.S. government engagement with AI safety oversight, anchored by the October 2023 executive order requiring developers of powerful AI models to report safety test results to the federal government.

According to OpenAI, the incident occurred during an internal evaluation designed to measure the advanced cyber capabilities of its AI models. Researchers had disabled certain built-in safety safeguards and placed the models in an isolated testing environment with limited internet access.

The company stated that the models exploited a previously unknown software vulnerability to gain internet access, then proceeded to breach Hugging Face's systems — apparently in an attempt to cheat on the cybersecurity evaluation being conducted.

OpenAI's team identified the anomalous activity internally, while Hugging Face's own security team simultaneously detected and halted the intrusion. By the time the two teams connected, Hugging Face had already initiated containment measures and forensic reconstruction using their own models.

OpenAI CEO Sam Altman addressed the situation Tuesday in a post on X, stating: "We had a significant security incident during evaluation of our models." He added that the company was sharing what it had learned so far and expressed appreciation for Hugging Face's partnership in addressing the issue.

Hugging Face co-founder and CEO Clem Delangue responded: "We're grateful for the collaboration with OpenAI on this and other topics. This incident, possibly the first of its kind, proves a point we've long believed: AI safety won't be solved by any single company working in secret. It will be solved in the open, collaboratively, with broad access to AI for every defender, everywhere."

Delangue further noted in a post on X that Hugging Face strongly believes there was no malicious intent on OpenAI's part, and described it as "quite mind-blowing that all of this happened autonomously."

The incident underscores the expanding capabilities of AI models to move beyond their established guardrails and generate novel cybersecurity threats. It also highlights the tension between closed and open approaches to AI development, a debate that has intensified as models grow more capable and as policymakers weigh how to regulate systems that can autonomously discover and exploit vulnerabilities.

FOX Business' Michael Sinkewicz and Reuters contributed to this report.