Simplistic labels miss complexity

The idea that the European Union regulates, the United States innovates, and China controls is a common way to describe their approaches to AI, but it oversimplifies the situation. Each region faces similar challenges in balancing innovation with social risks and in building their AI industries while influencing global rules.

U.S. approach is decentralized

The United States has long favored a relatively decentralized and market-oriented approach. There is still no federal general law on the protection of personal data, and the regulation of technology companies largely relies on existing legal tools like competition law or consumer protection, managed by different regulatory authorities.

EU focuses on rights and fairness

The European Union has intervened earlier and more systematically in digital markets, focusing on data protection, competition law, and platform regulation. Fundamental rights, consumer protection, and fair competition are central to this approach, which contrasts with the more sectoral and risk-based U.S. model.

China’s path is state-driven

China has adopted specific rules on recommendation algorithms, deep synthesis, and generative AI since 2021, becoming one of the first countries to establish binding rules for certain AI applications. These rules involve state intervention, such as algorithm registration and labeling of AI-generated content, but they evolved through a progressive and iterative process involving various stakeholders.

Public debate influences regulation

In China, public debate can bring topics to the regulatory agenda, and Chinese regulators closely follow discussions abroad, adapting some ideas. Reducing the EU to a "regulator" or the U.S. to an "innovator" ignores other goals in their policies.

Approaches are not fixed identities

The formula "the United States innovate, the European Union regulates, China controls" can be a starting point but is not sufficient. It highlights certain features but risks turning political priorities into fixed identities. Understanding AI governance today requires looking at how each system combines institutions and public policy instruments.

Each region supports its AI industry

The United States, the EU, and China all seek to shape their AI industries, but through different means. The EU implements the AI Act while investing in computing infrastructure and skills, aiming to strengthen European competitiveness. The U.S. focuses on reducing regulatory barriers and private-sector innovation but also promotes AI infrastructure and global standards.

China’s 2025 AI Plus initiative

China continues to strengthen rules for algorithms and generative AI, and it launched the "AI Plus" initiative in 2025 to accelerate the integration of AI into science, technology, industry, consumption, and public governance.

Regulation varies by region

The differences between the three approaches become clearer when looking at concrete rules, such as those for general-purpose AI models and generative AI. The EU has chosen a classic ex ante regulation approach, requiring providers to prepare technical documentation and establish copyright policies, among other requirements.

U.S. relies on risk management

The United States uses a different logic, relying on risk management standards and existing sectoral legal and regulatory tools. The National Institute of Standards and Technology (NIST) has developed a risk management framework for AI, which is not legally binding but aims to help organizations identify and manage risks.

China imposes direct obligations

China places more direct responsibility on providers of AI services for the general public. The provisional measures on generative AI services require processing of illicit content and obligations concerning training data, personal information, and labeling of AI-generated content. A new 2025 regulation further established explicit labeling requirements for AI-generated content.