The Bletchley Declaration

The UK's AI Safety Summit met on 1 and 2 November 2023 at Bletchley Park, where Turing had worked on the German ciphers, and its first-day statement was signed by 28 governments including the United States, China, the European Union, India and Brazil. The declaration recognised that frontier models could cause "serious, even catastrophic, harm, either deliberate or unintentional", agreed that the companies building them bore responsibility for their safety, and committed the signatories to meet again. Elon Musk interviewed Rishi Sunak on stage; Kamala Harris announced an American safety institute in a speech in London the same day.

The summit produced institutions. The UK's AI Safety Institute, later renamed the AI Security Institute, began testing frontier models before release under voluntary agreements with the laboratories; the US established its own; Japan, Singapore, Canada and others followed; and further summits met in Seoul in May 2024 and Paris in February 2025, where the American delegation declined to sign the communiqué.

Bletchley is on this timeline as the first time governments treated the technology as a matter of collective safety rather than industrial policy, and as the start of a two-year period in which that framing held before the competitive one returned.