The Seoul summit
South Korea and Britain held the second international AI safety summit; states adopted the Seoul Declaration, and companies for the first time promised public risk thresholds.
Why it matters
The conversation moved from whether rules were needed to which numbers companies would agree to say out loud.
Bletchley produced a joint statement of intent; Seoul added commitments with specifics. Companies developing frontier models agreed to publish safety frameworks with thresholds beyond which risk is treated as unacceptable, and to describe what they would do in that case. The declaration was backed by the European Union and a range of states, among them Korea, Britain, the United States, France, Germany, Italy, Japan, Canada, Australia and Singapore. No enforcement mechanism appeared, but a text against which actions can be checked did.