Trump administration disrupts Burnham's AI safety plans

UK Prime Minister Andy Burnham's plan to position Britain as a global leader in AI safety has been undermined by the Trump administration, which has delayed US AI firms from sharing models with the UK's AI Security Institute (AISI) and rejected international agreements. The White House told OpenAI and Anthropic not to share their models with the UK until US testers finish assessing threats to American systems. Burnham had hoped to use the UK's G20 presidency to set international AI principles, but now faces the challenge of either bringing the US onside or proceeding without its closest ally. The UK previously hosted the 2023 Bletchley Park AI summit but declined to sign a 2025 Paris declaration alongside the US. A 22-country UN pact on frontier AI safety this week excluded both nations, while China has shown greater willingness to cooperate on global AI standards.