A change of approach in Washington? This is the country with which AI giants will be allowed to share information
The Trump administration is allowing American AI developers, including OpenAI and Anthropic, to share new models with the British AI Safety Institute before their launch, after previously banning this. The decision, published on October 12, 2026, marks a shift in the president's policy, who had previously rejected strict international oversight. Now companies can transfer information without fear of violating White House guidelines.
The Trump administration is changing its policy and allowing American AI companies, including OpenAI and Anthropic, to share new models with the British AI Safety Institute before their launch. The decision was published on October 12, 2026, and marks a turnaround from the previous White House stance, which banned companies from transferring information to the institute. According to a report in The Telegraph, Washington has withdrawn a previous demand. Spokespeople for the president and the companies declined to comment. The British institute was established after the global safety summit in 2023 and examines advanced models for risk assessment. The American move expands cooperation with London, while the world tries to regulate AI oversight. Trump previously rejected calls to impose binding safety requirements, but on Friday the new task force he established determined that companies will be required to report security incidents.