AI Seoul Summit adopts the Seoul Declaration

Background

International meetings on AI give governments a place to put shared positions in writing. The AI Seoul Summit, held in May 2024, was one such meeting, and its output was a set of signed documents rather than a treaty. The summit took place in videoconference format at the state guesthouse Yeongbingwan in Seoul ^[2]. The leaders' session was co-chaired by the President of the Republic of Korea and the UK Prime Minister ^[2].

What was adopted

The summit produced three outcome documents: the Seoul Declaration, the Seoul Statement of Intent toward International Cooperation on AI Safety Science, and the Seoul Ministerial Statement ^[1]. They were signed over two consecutive days, and the lists of signatories differ in size. Ten countries and the European Union signed the Seoul Declaration on 21 May 2024 ^[1]. The next day, 22 May 2024, twenty-seven countries and the European Union signed the Seoul Ministerial Statement ^[1].

The leaders' session adopted the declaration together with its annex ^[2]. Participants welcomed the Seoul Statement of Intent toward International Cooperation on AI Safety Science as that annex to the declaration ^[1]. So the three documents are linked: a declaration, a statement of intent attached to it as an annex, and a separate ministerial statement with a longer list of signatories.

What the declaration says

The declaration noted the Frontier AI Safety Commitments and supported operationalisation of the Hiroshima Process International Code of Conduct for organisations developing advanced AI systems ^[1]. "Frontier AI" generally refers to the most capable general-purpose models available at a given moment; "operationalisation" means turning a written code into practices an organisation can actually follow, such as internal review steps or reporting routines. The declaration also looked forward to the third gathering at the AI Action Summit to be held in France ^[1].

Why it matters

Documents of this kind are political commitments, not treaties: governments record what they say they will do, and later meetings can point back to the text. The Seoul Declaration connects the summit's output to work already under way, naming the Frontier AI Safety Commitments and backing operationalisation of the Hiroshima Process International Code of Conduct ^[1]. Naming a next venue serves the same purpose: the declaration looked ahead to the AI Action Summit in France, described as the third gathering in the sequence ^[1].

What the documents do not do is set up an enforcement mechanism. They fix shared language, record which governments were willing to put their names to it, and leave the follow-through to the signatories. That is also why the signatory counts are worth noting: the declaration and the ministerial statement were signed by different-sized groups on consecutive days ^[1], so the record shows both the narrower group behind the leaders' text and the wider group behind the ministerial one.

The declaration's annex points in a second direction. By welcoming the Seoul Statement of Intent toward International Cooperation on AI Safety Science ^[1], participants put a document about the science of AI safety alongside the commitments made by developers. The summit thus left behind a short paper trail — a declaration, an annex on safety science, and a ministerial statement — plus a date and place for the next round, in France ^[1].