Page 23 - The Annual AI Governance Report 2025 Steering the Future of AI

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

Building Common Ground: Track 2 initiatives have produced consensus statements (e.g., the
Ditchley Statement, Beijing Dialogue 2024 ) that underline the need for coordinated global
51
52
action to prevent unacceptable AI risks, such as misuse and loss of control, and have informed
subsequent official (Track 1) negotiations. Dialogues have focused on developing best practices
53
for AI governance, including compute thresholds, model registration, and risk evaluation, with
technical experts from both global north and south institutions sharing approaches.
54

3.3 Regional AI Partnerships

Bilateral Agreements for AI Safety and Innovation: Bilateral agreements are quickly becoming
the responsive layer of AI safety governance. In April 2024, the US and UK AI Safety Institutes
signed the first formal partnership, pledging to carry out joint red-team exercises and share
evaluation datasets for frontier models. This model has since been adopted by other countries:
55
In November 2024, Singapore signed a Memorandum of Cooperation with the UK, and in
56
June 2025, Canada followed suit, linking its new Safety Institute and the firm Cohere to the
57
same testing protocols. At the strategic end of the spectrum, in November 2024, Washington
58
and Beijing agreed that only humans, not AI, will hold nuclear launch authority, highlighting a
shared interest in hard safety boundaries despite their rivalry. 59

Interregional Cooperation: There is growing momentum from bilateral agreements to
transregional initiatives. At their summit in Kuala Lumpur in May 2025, ASEAN, the Cooperation
Council for the Arab States of the Gulf (GCC) and China agreed to develop a joint framework for
data security cooperation and shared skills programmes. In December 2024, the EU and the
60
Smart Africa Alliance launched a Global Gateway partnership that pairs secure digital networks
with an African-owned data governance flagship to promote trustworthy AI across the continent.
61
The African Union’s Continental AI Strategy, endorsed in July 2024 and augmented in March
2025, sets continent-wide benchmarks on data sovereignty, trusted digital infrastructure and
62
ethical use of AI; by April 2025 AU technical assistance had enabled 22 member states to begin
drafting aligned national strategies.

51 Russell, S., Bengio, Y., Yao, A., Felten, E., Grosse, R., Hadfield, G., Khareghani, S., Hadfield-Menell, D., Perset,
K., Song, D., Zhang, Y.-Q., Chen, X., Tegmark, M., Seger, E., Zeng, Y., Zhang, H., He, Y.-H., Gleave, A., &
Heide, F. (2023, October 18–20). International Dialogue on AI Safety (Ditchley Statement). Ditchley Park, UK.
52 International Dialogues on AI Safety. (2024, March 10–11). Consensus Statement on Red Lines in Artificial
Intelligence. Beijing, China.
53 Siddiqui, S., Loke, K., Clare, S., Lu, M., Richardson, A., Ibrahim, L., McGlynn, C., Ding, J. (2025). Promising
topics for US–China dialogues on AI safety and governance.
54 Guest, O. (2025, April 11). International AI Safety Dialogues: Benefits, Risks, and Best Practices — Institute for
AI Policy and Strategy.
55 Alder, M. (2024, April 2). U.S., Britain announce partnership on AI safety, testing. Reuters.
56 Ministry of Digital Development and Information, Singapore; & Prime Minister’s Office, UK. (2024, November
6). New Singapore–UK agreement to strengthen global AI safety and governance. Government of Singapore.
57 Prime Minister’s Office, UK & Prime Minister’s Office, Canada. (2025, June 15). Joint statement between the
Prime Minister of the United Kingdom and the Prime Minister of Canada. GOV.UK.
58 Cohere Labs. (2025, June 15). Cohere partners with Canada and UK Governments on secure AI. Cohere.
59 Alder, M. (2024, November 17). Biden, Xi agree that humans, not AI, should control nuclear arms. Reuters.
60 Association of Southeast Asian Nations; Cooperation Council for the Arab States of the Gulf; & People’s
Republic of China. (2025, May 27). Final Joint Statement of the ASEAN-GCC-China Summit, Kuala Lumpur,
Malaysia.
61 Directorate-General for International Partnerships (European Commission). (2024, December 3). Global
Gateway: EU and Smart Africa strengthen partnership for Africa’s digital transformation. European
Commission.
62 African Union, (2024, August 9) Continental Artificial Intelligence Strategy, African Union.

14