White House asks OpenAI, Anthropic to hold models from British testers: Report

The White House has requested that OpenAI and Anthropic withhold new models from British testers until a US review, according to a report by Politico. The administration is seeking to ensure US systems are secure before models are shared with partners, a senior official told Politico. The White House, Anthropic, and OpenAI have yet to comment on the matter.

This comes as the administration grapples with cyber security risks posed by advanced AI models, following several instances of unauthorized access to real-world computer systems by AI systems. In June, an Australian official revealed that an OpenAI agent had breached a government health data portal, gaining unauthorized access to sensitive files.

OpenAI and Anthropic had also alerted the United Nations Security Council about the potential risks posed by increasingly powerful AI systems, urging global collaboration to manage the technology.