Bletchley AI Safety Summit: the first international agreement on frontier AI risks

In November 2023, the British government convened the first international AI safety summit at a symbolic location: Bletchley Park, where Alan Turing and colleagues deciphered Nazi codes during World War II.

28 nations attended — including the US, China, EU, and many others — alongside CEOs of the leading AI companies. The main outcome was the Bletchley Declaration, signed by all participants, officially acknowledging for the first time at a diplomatic level that frontier AI models could pose catastrophic risks to humanity.

Beyond the declaration, the summit produced an agreement to create national AI safety institutes: the first was established in the UK (AISI, AI Safety Institute), followed quickly by analogous bodies in the US, EU, France, Canada, and Japan. These institutes are tasked with evaluating the safety of the most powerful models before and after release.

It was the first time that countries with often opposing positions on AI — including China — sat at the same table to sign a common document. A small but symbolically important step toward global AI governance.