AI Safety Summit

In short

The AI Safety Summit was the first global government meeting on the risks of advanced AI, held by the UK at Bletchley Park in November 2023. It started a series of summits, later held in Seoul, Paris and New Delhi.

Also called: Bletchley Park summit, Bletchley Declaration, AI Seoul Summit, AI Action Summit, AI Impact Summit
Topic: Law, policy and the UK
Level: Beginner
First described: Hosted by the UK government at Bletchley Park on 1 and 2 November 2023, producing the Bletchley Declaration¹.

The first summit ran on 1 and 2 November 2023 at Bletchley Park. It produced the Bletchley Declaration, signed by 28 countries and the European Union, including the US and China¹.

The declaration focused on "frontier AI", the most capable general-purpose models. It said there is "potential for serious, even catastrophic, harm, either deliberate or unintentional" from their most significant capabilities, and the signatories agreed to meet again in 2024¹. The UK's AI Safety Institute was set up around the same time⁵.

The second meeting, the AI Seoul Summit, was co-hosted by the UK and South Korea in May 2024. There, 16 AI companies signed the Frontier AI Safety Commitments, promising to publish safety frameworks².

The series then moved away from "safety" in its name. France hosted the AI Action Summit in Paris in February 2025. Its main statement, on "inclusive and sustainable" AI, was not signed by the UK or the US³. Downing Street said the text lacked practical clarity on global governance and did not do enough on national security⁶.

India hosted the AI Impact Summit in New Delhi in February 2026. India's IT ministry said 91 countries and organisations had joined its New Delhi Declaration by 24 February⁴.

None of these declarations is legally binding. They are political statements of shared concern and intent.

Example

At Bletchley Park in November 2023, 28 countries and the EU, including both the US and China, put their names to the same statement on frontier AI risk¹.

Why it matters

The summits are where governments and AI companies make shared promises about advanced AI. Who signs, and who refuses, shows how far countries agree on the risks.

Questions people ask

What is the Bletchley Declaration?

A statement agreed at the November 2023 AI Safety Summit by 28 countries and the EU. It recognises that frontier AI could cause "serious, even catastrophic, harm" and commits signatories to work together on those risks. It is not legally binding¹.

Did the UK sign the Paris AI summit declaration?

No. The UK and US did not sign the February 2025 Statement on Inclusive and Sustainable Artificial Intelligence³. The Prime Minister's spokesperson said it lacked practical clarity on global governance and did not address national security enough⁶.

Where was the AI summit in 2026?

In New Delhi, India, in February 2026, under the name AI Impact Summit. By 24 February, 91 countries and organisations had joined its declaration, according to India's IT ministry⁴.

Related terms

Sources

Back to the glossary