AI Safety Summit 2024
- May 21, 2024
AI Safety Summit
The AI Safety Summits are bi-annual (approximately every six months) international meetings aimed at discussing the safety, governance and regulation of advanced AI systems — especially so-called “frontier” AI.
Agenda & Discussion Streams
While the full minute-by-minute agenda is not widely published in one place, major structural elements include:
A leaders’ session (heads of state/government) at which the “Seoul Declaration for Safe, Innovative and Inclusive AI was agreed.
Ministerial/wider stakeholder sessions on the day(s) around themes of safety, innovation and inclusivity.
Engagement with major AI companies: the announcement of the Frontier AI Safety Commitments by leading firms.
Announcement of an international network of AI safety institutes.
Purpose:
To build international consensus and practical cooperation on how to develop, deploy, and govern artificial intelligence safely, while preserving innovation and inclusivity.
Key Documents & Declarations
- Seoul Declaration for Safe, Innovative and Inclusive AI – adopted by participating leaders.
- Seoul Statement of Intent toward International Cooperation on AI Safety Science (Annex to the Declaration) – outlines scientific cooperation and network of safety institutes.
- Frontier AI Safety Commitments – leading AI organisations commit to thresholds, transparency, risk mitigation before deployment of “frontier” models.
Co-hosted by
South Korea & United Kingdom.
Summit Information
About the Summit
Major Outcomes & Commitments
- Agreement by 10 + countries plus EU to create an international network of AI safety institutes to collaborate on research, model testing, monitoring harms, sharing data.
- Adoption of voluntary commitments by major AI companies to halt development/deployment of models if risks from “frontier AI” cannot be mitigated below certain thresholds.
- The Seoul Declaration emphasised human-centric AI, democracy, human rights, rule of law, bridging divides.
- Governments reinforced that risks from AI include misuse, loss of oversight, societal integration issues, and that action must proceed alongside innovation.