AI Safety Institutes: Characteristics, Functions, and Challenges to date globally

The Institute for AI Policy and Strategy has produced a report titled “Understanding the First Wave of AI Safety Institutes: Characteristics, Functions, and Challenges” (here). 

The report looks at the first wave of AI Safety Institutes (AISIs) in UK, US and Japan.

These bodies were all heavily influenced by the process in the UK that culminated with the Bletchley summit, and share a set of common characteristics and functions that revolve around safety evaluations of advanced AI systems

In summary, the report identifies in the first wave of AISIs the:

• characteristics - they are technical government institutions; they have a clear mandate related to the safety of advanced AI systems; they have no regulatory powers.
• functions - a combination of research, standards and co-operation, with AI model evaluations at the centre;
• challenges -  according to the report, some analysts have criticised the first wave of AISIs for specializing too much, overlapping with existing institutions, such as existing standards-developing bodies, and their relationship with industry with perceived risk to impartiality. 

The report also compares the first wave AISIs with other AI institutes, such as the EU AI Office, created under the EU AI Act, and those in Singapore, Canada and France. These share similarities with AISI but may represent different models emerging, covering additional areas such as innovation.

If you would like to discuss how current or future regulations impact what you do with AI, please contact Tom Whittaker, Brian Wong, Lucy Pegler, David Varney, Martin Cook or any other member in our Technology team.

For the latest on AI law and regulation, see our blog and sign-up to our AI newsletter.