Trump reverses stance, lets UK access unreleased AI models again
What's the story
The UK will once again have access to advanced US artificial intelligence (AI) models before their public release. The decision comes after the White House reversed its earlier stance, which had blocked transatlantic cooperation on AI safety. The Trump administration had previously asked top developers to keep unreleased systems from Britain's AI Security Institute (AISI).
Review process
Model access part of broader AI safety debate
The policy change will enable UK officials to review models from OpenAI, Anthropic, and DeepMind before they are released to the public.
The decision comes after lobbying efforts by British ministers, including AI Minister Kanishka Narayan.
The move is part of a wider debate on how governments should evaluate increasingly powerful AI systems without stifling innovation.
Institute's function
AISI's role in AI safety
The AISI was set up after the 2023 AI Safety Summit, hosted by the UK. Its main job is to assess the capabilities of advanced AI systems and look for potential harmful use.
The institute also assesses risks that could arise before a model is released to the wider public.
Early access is critical, as it allows researchers to test systems under controlled conditions before their capabilities are widely available.
Access conflict
Dispute over access to models last month
The dispute over access first emerged last month when the White House reportedly told major American AI companies not to give the institute access to new models before release.
Anthropic then blocked British officials from accessing its Claude Mythos 5.1 model, The Telegraph reported.
The restoration of access is a reversal of a policy that could have limited the institute's ability to evaluate some of the most advanced systems being developed in the US.
Regulation hurdles
International standards for AI oversight challenged
The episode highlights the challenges of creating unified international standards for AI oversight when governments have different views on innovation and regulation.
President Donald Trump has long opposed mandatory safety requirements for AI developers.
British officials, on the other hand, have pushed for more international coordination and plan to use their G20 presidency next year to push talks on shared principles and standards for advanced AI.
Defense measures
UK government moves to strengthen defenses against AI threats
The renewed access comes as British ministers plan a series of measures to strengthen the country's defenses against AI-related threats.
Britain's AI minister Kanishka Narayan is expected to argue that tech companies shouldn't be solely responsible for managing risks from their own systems.
Proposed amendments to Labour's Cyber Security and Resilience Bill would increase reporting requirements for organizations involved in critical national infrastructure when AI-related security incidents occur.
Cybersecurity strategy
Advanced AI tools for cybersecurity defense
The UK government is also considering the wider use of advanced AI tools to protect Whitehall departments from cyber attacks.
More resources are expected to be allocated for officials at the National Cyber Security Centre, the Cabinet Office, and the Department for Culture, Media and Sport to investigate incidents and improve responses.
These measures come after warnings from MI6, MI5, and GCHQ heads that AI could give hostile states new ways to target sensitive infrastructure or obtain confidential information.