The White House has directed leading artificial intelligence developers to allow United States officials to review new models before they are shared with British testing partners. The move places a domestic gatekeeping step ahead of international collaboration for frontier AI systems.
What Happened
The Decoder reports that the administration has instructed OpenAI and Anthropic to submit their latest model iterations for US government review prior to releasing them to testers in the United Kingdom. This directive effectively inserts a national security or policy assessment phase into the pre-release workflow for these major AI labs.
Why It Matters
The order signals an increasing emphasis on sovereign control over frontier AI capabilities. By mandating a US review before international sharing, the government places a domestic assessment phase ahead of foreign interaction with the most advanced models.
The Bottom Line
US authorities are tightening oversight of how frontier AI models are shared internationally, requiring a pre-review step for OpenAI and Anthropic before British testers can access new versions.