The Gist
The United Nations has developed a framework for AI governance focused on its ethical use and societal impact but has neglected the actual production and evaluation of AI technologies. This oversight creates a significant gap in global AI safety and governance.
How It Worked
While the UN engages on the demand side of AI technology, countries like the UK and the US have established voluntary agreements for pre-deployment evaluations of AI models with major labs. This model demonstrates a pathway for the UN to adopt a similar role, ensuring that all countries can benefit from safety assurances and oversight. The UN can leverage existing structures, such as the International Computing Center, to create a small standing-evaluation unit for AI.
Results
Currently, the U.S. has authorized Anthropic's AI model for about 100 American institutions, while many other countries remain without any oversight framework. This selective release highlights a disparity in access to advanced AI technologies, leaving 191 countries without the same safety evaluations that only two countries currently enjoy.
Why It Matters for You
For businesses and governments globally, understanding the importance of a multilateral approach to AI oversight is crucial. Engaging in dialogue about AI governance can lead to more equitable access to technology and shared safety standards. As the AI landscape evolves, advocating for a UN role in supply-side oversight can help ensure that all nations benefit from advancements in AI technology responsibly.
