Bletchley Park AI Safety Summit ends with a 28-country declaration
The UK-hosted summit closed on 2 November 2023 with 28 countries and the EU signing the Bletchley Declaration on risks from frontier AI.
This summary was written by AI to complete the context; it is not based on a current source.
The AI Safety Summit was held at Bletchley Park, the wartime codebreaking site north of London, on 1 and 2 November 2023. It was hosted by the British government under Prime Minister Rishi Sunak and was the first major international gathering devoted to the risks of advanced artificial intelligence.
At its close, 28 countries and the European Union signed the Bletchley Declaration. The signatories included the United States and China, which rarely align on technology policy. The text recognised that the most capable "frontier" AI models could pose serious risks, including in areas such as cybersecurity and biotechnology, and committed the parties to cooperate on understanding and managing them.
The declaration was non-binding and set out no enforcement rules. Even so, it mattered as a first shared statement of concern among major AI powers. The summit also led to agreements by leading AI companies to let governments test models before release, and to the creation of AI safety institutes, starting with one in the UK. It also set up follow-up summits, with South Korea and then France taking up the series.
Story timeline
Paris AI summit on AI safety regulation · 3 updates
What happened before and after this story. Background events not previously covered were summarised by AI.



