Page 22 - The Annual AI Governance Report 2025 Steering the Future of AI

The Annual AI Governance Report 2025: Steering the Future of AI

Theme 3: International AI Governance Coordination

3.1 Global AI Summits

Formalizing Global AI Summit Series: The Global AI Summit Series is a major international International AI Theme 3:
gathering bringing together governments, industry, and civil society to establish AI norms. The
Global AI Summit Series, launched at Bletchley Park (1-2 November 2023) , continued in Seoul
42
(21-22 May 2024) and France AI Action Summit 44,45 (10-11 February 2025), establishing itself
43
as the central venue for international collaboration on AI governance, producing foundational
documents like the Bletchley Declaration and the Seoul Declaration. These summits have
46
47
shifted from ad hoc gatherings to a more structured, recurring process, with recommendations
to further institutionalize the series for sustained impact and clearer agenda-setting. India will
host the next summit on 19-20 February 2026.

Advancing Concrete Commitments: At the AI Seoul Summit (21–22 May 2024), sixteen frontier-
model developers, including Amazon, Anthropic, Google, Microsoft, and Samsung, signed the
Frontier AI Safety Commitments, pledging to conduct red-team testing and implement model-
weight security measures. They also committed to publishing public "severe-risk" frameworks
48
ahead of the Paris AI Action Summit in February 2025. A civil society Seoul Commitment
Tracker released after the 10 February 2025 deadline found that only six firms had fully met
the commitments, four were partially compliant, and six had fallen short, highlighting uneven
corporate engagement.
49

3.2 Track 2 Diplomacy Initiatives

Track 2 Diplomacy in AI Governance: Track 2 diplomacy refers to informal, non-government
channels—such as academics, former government officials or civil society leaders—who meet
to explore ideas and issues that their governments (the “Track 1” diplomats) may be unable or
unwilling to discuss in public. A series of recurring Track 2 dialogues between US and Chinese
experts—such as those held in Thailand (2024) and organized by institutions like Tsinghua
University and Brookings—have become essential forums for candid, technical discussions on
AI safety, military AI, and risk mitigation, especially where official channels are constrained.
50
These dialogues have addressed sensitive scenarios, using simulations and tabletop exercises
to move from abstract principles to practical risk management.

42 UK Prime Minister’s Office. (2023, June 7). UK to host first global summit on Artificial Intelligence [Press
release]. GOV.UK. Retrieved June 25, 2025.
43 About the AI Seoul Summit 2024. (2024, May 7). GOV.UK.
44 Présidence de la République (France). (2025, March 13). Sommet pour l’action sur l’intelligence artificielle
[Web page]. Présidence de la République française. Retrieved June 25, 2025.
45 Dennis et al. (2025). What success looks like for the French AI Action Summit. Centre for the Governance of
AI.
46 UK Prime Minister’s Office, Foreign, Commonwealth & Development Office, & Department for Science,
Innovation and Technology. (2025, February 13). The Bletchley Declaration by countries attending the AI
Safety Summit, 1–2 November 2023 (Policy paper). GOV.UK. Retrieved June 25, 2025.
47 Ministry of Foreign Affairs, Republic of Korea. (2024, May 23). Seoul Declaration for safe, innovative and
inclusive AI by participants attending the leaders’ session of the AI Seoul Summit, 21 May 2024 [Press release].
Retrieved June 25, 2025.
48 Department for Science, Innovation and Technology. (2025, February 7). Frontier AI Safety Commitments,
AI Seoul Summit 2024. GOV.UK.
49 Seoul Commitment Tracker.
50 Kahl, C., & Hass, R. (2024, April 5). Laying the groundwork for US-China AI dialogue. Brookings.

13