AdobeStock © Supatman
On 20 and 21 November 2024, AI safety institutes and government-mandated offices from Australia, Canada, the European Commission, France, Japan, Kenya, the Republic of Korea, Singapore, the United Kingdom, and the United States are convening in San Francisco for the first meeting of the International Network of AI Safety Institutes.
Building on the Seoul Statement of Intent toward International Cooperation on AI Safety Science, released during the AI Seoul Summit on 21 May 2024, this initiative marks the beginning of a new phase of international collaboration on AI safety.
The Network brings together technical organisations dedicated to advancing AI safety, helping governments and societies better understand the risks posed by advanced AI systems, and proposing solutions to mitigate these risks. The Network members also stress in their Mission Statement that “international cooperation to promote AI safety, security, inclusivity, and trust is vital to addressing these risks, driving responsible innovation, and expanding access to the benefits of AI worldwide.”
Beyond addressing potential harms, the institutes and offices involved will guide the responsible development and deployment of AI systems.
Goals and priorities of the network
The International Network of AI Safety Institutes will serve as a forum for collaboration, bringing together technical expertise to address AI safety risks and best practices. Recognising the importance of cultural and linguistic diversity, the Network will work towards a unified understanding of AI safety risks and mitigation strategies.
It will focus on four priority areas:
- Research: Collaborating with the scientific community to advance research on the risks and capabilities of advanced AI systems, while sharing key findings to strengthen the science of AI safety.
- Testing: Developing and sharing best practices for testing advanced AI systems, including conducting joint testing exercises and exchanging insights from domestic evaluations, as appropriate.
- Guidance: Facilitating shared approaches to interpreting test results for advanced AI systems to ensure consistent and effective responses.
- Inclusion: Engaging partners and stakeholders in regions at all stages of development, by sharing information and technical tools in accessible ways to broaden participation in AI safety science.
A commitment to global cooperation
Through this Network, the members commit to advancing international alignment on AI safety research, testing, and guidance. By fostering technical collaboration and inclusivity, they aim to ensure that the benefits of safe, secure, and trustworthy AI innovation are shared widely, enabling humanity to fully realise AI’s potential.