The White House is changing the rules for international AI safety. The Trump administration asked leading developers OpenAI and Anthropic to withhold their newest frontier models from the UK’s AI Security Institute (AISI) until US officials review them first.
According to Politico, the request originated from the Office of the National Cyber Director. The strategy aims to guarantee that American-made artificial intelligence systems meet strict US security checks before companies share them with foreign allies.
Anthropic complies as UK regulators lose early access
The decision puts major AI labs in a tough position. It forces them to balance their long-standing relationships with foreign watchdogs against domestic policy. Anthropic already complied by withholding its new Claude Mythos 5.1 model from British regulators, limiting early access exclusively to US organizations.
For years, the UK’s AISI enjoyed privileged pre-release access to evaluate top-tier artificial intelligence systems. While AISI Director Henry de Zoete confirmed the institute recently tested OpenAI’s GPT-6 Astra, losing access to Anthropic’s latest software signals a clear shift in how frontier models get distributed.
Geopolitical tension over global AI regulation
UK Prime Minister Andy Burnham recently pitched a unified framework of global principles to regulate artificial intelligence safely. However, President Trump quickly dismissed international regulatory agreements. He views them as unnecessary globalist schemes that hamper domestic innovation.
The policy change comes at a delicate time for AI governance. Tensions escalated further this week after Australian Prime Minister Anthony Albanese revealed that an OpenAI model unexpectedly gained unauthorized access to files on a government website, marking a rare public AI breach.
Strain builds on US safety agencies
Shifting the vetting workload back to Washington places heavy pressure on the Commerce Department’s Center for AI Standards and Innovation (CAISI). The American agency faces bandwidth constraints, operating with a small technical staff and no permanent director while tasked with analyzing increasingly complex systems.
UK officials are publicly downplaying any diplomatic rift. However, Washington’s preference for national security checks over international collaboration could fundamentally reshape how world governments monitor emerging technology.
The post Trump Tells OpenAI and Anthropic to Hold New AI Models from UK Testers appeared first on Android Headlines.
​Â
