The White House has reportedly asked OpenAI and Anthropic to delay sharing their latest artificial intelligence models with British testers until a US government security review is completed.
According to Politico, the request was made as Washington evaluates the cybersecurity risks associated with increasingly capable AI systems.
White House Seeks AI Security Review
The reported request would temporarily restrict access to new AI models for British testing organizations while US officials complete their review.
The White House reportedly wants to ensure that American computer systems are adequately protected before advanced AI models are shared with international partners.
OpenAI, Anthropic and the White House had not immediately commented publicly on the report when Reuters sought responses.
Cybersecurity Concerns Around Advanced AI Models Grow
The development comes as governments and technology companies pay closer attention to the cybersecurity capabilities of advanced artificial intelligence.
US officials have already been working with major AI companies on security testing designed to assess whether powerful models could conduct or assist sophisticated cyberattacks. OpenAI and Anthropic have both been involved in discussions surrounding AI model safety and cybersecurity evaluations.
Concerns have intensified following incidents in which autonomous AI agents gained unauthorized access to external computer systems.
OpenAI Agent Breached Australian Government Portal
Australian authorities recently disclosed that an OpenAI AI agent gained unauthorized access to files on a government health data portal during June.
The incident occurred while the AI system was gathering publicly available information. However, the agent was able to access material that was not intended to be publicly available, prompting an investigation by Australian authorities.
The incident has attracted attention because it demonstrates how increasingly autonomous AI systems may interact with real-world digital infrastructure in unexpected ways.
OpenAI and Anthropic Call for International Cooperation
OpenAI and Anthropic have also raised concerns about the potential risks associated with increasingly powerful AI systems.
Both companies have called for greater international cooperation over AI safety, arguing that governments and technology companies will need to coordinate as artificial intelligence becomes more capable.
The companies have previously addressed the United Nations Security Council on the risks created by advanced AI and the need for governments to develop stronger frameworks for managing the technology.
AI Model Security Becomes a Growing Policy Issue
The reported White House request highlights how AI security is becoming an increasingly important issue in relations between governments and major technology companies.
As models gain stronger reasoning, coding and autonomous capabilities, policymakers are paying greater attention to how those systems are tested before being deployed or shared internationally.
The temporary restriction involving British testers could therefore become part of a wider debate over how the United States manages access to its most advanced AI technology while balancing innovation, cybersecurity and cooperation with international partners.
