National AI Policy Trends Around the World - Disruptive Competition Project (2025)

Countries are rightly recognizing the powerful role artificial intelligence (AI) can play in achieving their national agendas. Machine learning advancement and computational power increases have fueled rapid development, leaving governments eager to stay abreast of technological advancements and implement policies that will keep pace. However, as AI promotion and regulation are at the forefront of policymakers’ minds, lawmakers’ approaches from different countries are vastly different. These differences in turn increase the risk that businesses will face fractured and inconsistent AI regulation. Nevertheless, certain overarching trends have emerged from the enacted AI policies.

Countries adopting hard law often follow a risk-based or sector specific approach.

A growing number of countries have enacted binding legislation, which establishes clear obligations and penalties for AI use and development. These laws can take a variety of forms, but the most common two types are a risk-based horizontal law or sectoral specific laws targeting specific industries.

The European Union adopted one of the world’s first risk-based approaches to AI. This approach categorizes AI systems based on their potential risks and tailors regulatory requirements according to the risk level. In December 2024, South Korea made waves by adopting a similar risk-based approach in their comprehensive AI legislation. The law subjects higher-risk AI systems to more rigorous regulatory obligations to mitigate the risks of greater potential harms. Other countries are expected to follow the precedent set by the EU and South Korea in the risk-based approach.

When countries are apprehensive to adopt a comprehensive law, they often begin by legislating certain industries that they deem need regulation. For example, countries like the United States, the United Kingdom, and France have all engaged in sector specific regulation for medical devices and autonomous vehicles. With AI, governments tend to favor sectoral-specific regulation when looking for precise control in areas where the technology has unique risks.

Countries adopting soft law often follow a decentralized approach and rely on overlap with other laws.

Most countries with an interest in AI have adopted or are working to create some form of a non-binding national AI plan or policy that recognizes the role this technology can play in achieving their national goals. These policies often identify voluntary AI principles and ethical guidelines, which can serve as a stop-gap measure while countries wait and see what to do next.

Countries also seek to update or amend existing laws and bodies to regulate AI. Whether charging existing federal agencies to update their policies or turning to existing data privacy laws, many countries are relying on structures already in place to navigate AI challenges. This approach allows countries to leverage existing enforcement expertise, infrastructure, and resources, especially in areas with a strong connection to AI like privacy and intellectual property.

Countries hoping to become a leader in AI often establish regulatory sandboxes.

A final emerging trend is the establishment of regulatory sandboxes for those countries looking to get ahead in AI development. Previously utilized by the financial sector, these controlled environments allow for experimentation with new technologies under regulator supervision, without exposing the public to unchecked risks. Sandboxes can be incredibly resource-intensive, but allow collaboration between industry, academia, and regulators.

For more information on national AI policies, see here.