OpenAI institutes new safeguards after Hugging Face breach

OpenAI has announced a new batch of security policies focused on containing incidents while models are being tested, weeks after a breach at Hugging Face saw models escape their training environment. The new safeguards include more detailed monitoring of models during the development process, as well as greater emphasis on alignment and security during post-training, the company said in a blog post. “As models become more capable, the risks associated with developing and testing them internally also grow,” OpenAI wrote. “Our standards for monitoring, alignment, and security must stay ahead of those risks.” The measures also include stronger network isolation: under the new system, a single compromise of a workload or supporting service does not, by itself, allow unauthorized access to the internet or other internal networks. The strongest safeguard is a monitoring system that examines tool actions, available reasoning traces, and activity logs for unauthorized behavior, with a stated goal of issuing alerts within 30 minutes of concerning activity. OpenAI estimates the monitoring will add roughly 20% compute overhead to whatever process is being monitored. OpenAI representatives said the policies are not a direct response to the Hugging Face incident, but were also provoked in part by the cybersecurity capabilities of the forthcoming Astra model and the overall pace of AI progress. The company also disclosed that it froze reinforcement learning for two weeks after the incident and has since restarted many of the less risky models, while its largest planned frontier RL run remains on hold. Speaking to reporters, OpenAI's VP of research Amelia Glaese said the strictness of the controls would increase as models become more capable, with the largest models facing the greatest scrutiny. OpenAI's official post-mortem of the incident is still pending, and the company has promised further details on the monitoring system in a forthcoming blog post. The announcement marks one of the first public changes in OpenAI's safety practices since the incident and signals how seriously frontier labs now take containment risks. The pending post-mortem — and the launch of Astra, which helped provoke the new policies — will be the first real tests of whether the new controls hold.