The passing of the European Union’s Artificial Intelligence Act (AI Act) in 2024 marked a key milestone in the governance of AI. The Act is one of the first general legal frameworks on AI anywhere in the world, and some believe it will shape the future of AI regulation globally. It establishes obligations for market actors based on the potential risks and level of impact posed by different AI systems. For “high-risk” systems, a range of mandatory requirements (including a conformity assessment) must be met before a system can be placed on the market. The Act also places specific obligations on providers of general-purpose AI models (also known as foundation models). 

In the United States, policymaking on AI has developed from a relatively hands-off agency-by-agency approach towards a more coherent regime. In 2023, President Biden issued an Executive Order on Safe, Secure and Trustworthy AI. The Order sets out requirements for AI transparency, testing, and cybersecurity measures, and directs a major programme of work across the federal government. Alongside White House initiatives, regulators such as the Federal Trade Commission (FTC) have shown increasing willingness to investigate and intervene in AI markets. Nevertheless, Congress is yet to pass any AI-specific legislation, and we are likely to see more state-level policymaking on AI in the years to come. 

Elsewhere around the world, other governments have proposed comprehensive legislation on AI. Canada's draft AI and Data Act follows a similar risk-based approach to that of the European Union, and Brazil has published an AI Bill that would establish a regulatory regime, including civil liability for AI providers. These efforts stand in contrast to more voluntarist approaches taken by countries like Singapore, which has developed various initiatives for responsible AI deployment.  

Beyond national level regimes, there is a growing focus on international cooperation on AI governance. In October 2023, under what is known as the Hiroshima Process, G7 leaders agreed Guiding Principles and a Code of Conduct for organisations developing the most advanced AI systems, including foundation models and generative AI systems. In November 2023, the UK government brought together governments and some leading AI companies to set the agenda for future international cooperation on AI safety. The Bletchley Declaration was signed by all countries attending the summit, including China and the United States.

Over the next few years, decision makers will determine the institutional arrangements that will govern AI globally. Some experts suggest that rather than a single institution, we will likely see the emergence of a regime complex comprising multiple institutions within and across several functional areas.  

Global Policy Venues

Bletchley Declaration 

In 2023, the UK government brought together actors from across governments, leading AI companies, civil society, and academia to set the agenda for future international cooperation on AI safety. The Bletchley Declaration was signed by all countries attending the summit, including China and the United States. A second AI Safety Summit was held online in May 2024, and will be followed by a further AI Action Summit to be held in France in February 2025. 

United Nations 

In October 2023, the United Nations Secretary-General António Guterres announced the creation of a High-level Advisory Body on AI to analyse and advance recommendations for the international governance of AI. The Advisory Body will publish its final report ahead of the Summit of the Future to be held in 2024.

Within the UN system, the ITU and UNESCO have recently led in inter-agency working group on AI. In 2024, the working group published a White Paper analysing the UN system's institutional models, functions, and existing international normative frameworks applicable to global AI governance. 

OECD

The Organisation for Economic Co-operation and Development (OECD) continues to be a key forum for the development of AI governance and assurance. In recent months, there has been considerable convergence around OECD definitions and benchmarks, and nominally risk-based regimes. This includes the updated definition of an AI system (2023):

“An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.”

G7 and the Hiroshima Process 

In April 2023, Japan hosted a meeting of the G7 Digital and Technology Ministers meeting where leaders agreed to prioritise collaboration for inclusive governance of the most advanced AI technologies. In October 2023, they agreed on the Hiroshima Process International Guiding Principles for Organizations Developing Advanced AI Systems, and the Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems. Both the Principles and the Code of Conduct will provide guidance for organisations developing and deploying the most advanced AI systems, including foundation models and generative AI systems.