AI Safety Institutes

Methodological note

An AI Safety Institute (AISI) is a government-backed institution dedicated to fostering the safe and secure development and deployment of AI. The International Network of AI Safety Institutes was established in November 2024, on behalf of the AI safety institutes and government mandated offices that facilitate AI safety and evaluation from Australia, Canada, the European Commission, France, Japan, Kenya, Korea, Singapore, the United Kingdom, and the United States. The mission statement of the Network is as follows:

“Our institutes and offices are technical organisations that aim to advance AI safety, help governments and society understand the risks posed by advanced AI systems, and suggest solutions to address those risks in order to minimize harm. Beyond mitigating risks, these institutes and offices play a crucial role in guiding the responsible development and deployment of AI systems.”

Common instruments (e.g. EU-level initiatives) are not taken into account. The rationale behind this choice is that including supranational efforts would not accurately reflect which individual countries are taking national leadership and investing directly in the institutional infrastructure needed to ensure AI safety.

For the OECD.AI Index, before the establishment of the Network, manual research was conducted to provide a score of 1 for countries with an established AI Safety Institute and 0 to those without. In 2024 and in subsequent iterations, a score of 1 is given to members of the Network, and 0 for non-members.

Some of the data underlying this indicator have been compiled through manual desk research. While OECD.AI endeavours to ensure the accuracy, consistency and methodological rigour of the data, errors or omissions may occur. The data is provided for informational purposes and should be interpreted accordingly.

If you identify any errors, omissions or potential corrections, please contact AI@oecd.org.