The Battle of the Rules — Regulation and Governance

Between the end of 2022 and the course of 2023, generative AI poured into the world as if a dam had broken. Anyone could hold in their hands a tool for producing text and images, and companies rushed to fold it into their own services. Progress in the technology no longer belonged to researchers alone.

And so a question that had until then been set aside suddenly stepped into the leading role. Who would bind this powerful tool, and by what rules? This “intelligence” was neither a weapon nor a medicine, so how was society to govern it at all? Facing a question with no ready answer, the world’s major powers each began to grapple with it from the standpoint of their own philosophy. This was the beginning of a contest not over the speed of technology, but over values.

Europe — Sorting by Risk and Drawing the Line

The first to move, and the most systematic, was the European Union. The EU proposed comprehensive legislation to regulate AI in 2021, and after a long debate brought it to fruition in 2024 as the AI Act. The European Parliament passed it in March 2024, and the law entered into force on 1 August 2024.

At the heart of the EU AI Act lies the idea of classifying AI according to risk. Uses that threaten people’s rights, such as social credit scoring, are treated as “unacceptable risk” and banned in principle; uses with grave consequences, such as in medicine or hiring, are treated as “high risk” and burdened with strict obligations. Meanwhile, the great majority of AI, such as the sorting of spam, is left largely unregulated and free. It is a graduated design, varying the strength of the constraint with the size of the risk.

Yet criticism clung to this path as well. If the regulation were too heavy, would Europe’s startups not be left unable to move, falling behind the United States and China in technological development? To prioritize safety, or to prioritize speed? The EU AI Act was also a grand experiment that carried that very tension within it. The law is designed to apply in stages, with the provisions on prohibited uses coming first and full application said to follow only later.

The United States — The Decree, and the Reversal

Across the Atlantic, the path taken by the United States, which runs at the very front of AI development, stood in contrast to Europe’s. Rather than crafting comprehensive legislation all at once, it chose first to act through executive command.

On 30 October 2023, then-President Biden signed an executive order titled “Safe, Secure, and Trustworthy AI” (Executive Order 14110). It was described at the time as the most comprehensive piece of AI governance ever undertaken by the U.S. government, a sweeping document directing dozens of federal agencies to take more than a hundred concrete actions. It showed a posture that reached into security, privacy, and fairness, including a requirement that companies developing powerful AI models report the results of safety testing.

But this command did not last long. When the administration changed, in January 2025 the new administration was reported to have revoked the Biden administration’s AI executive order. The policy that followed is said to have shifted toward removing regulation, accelerating AI development, and preserving U.S. advantage, rather than oversight and correction; views on it differ. Driven forward by a single decree and overturned by a single decree — the American style of moving through executive orders rather than laws carried, on its other face, a corresponding tendency for policy to waver.

Concern for safety, and the speed of development. Within the United States too, these two values went on contending with each other.

China — Governing the Ideas Themselves

The third pole, China, took an approach distinct from either of those in the West. In July 2023, several authorities including the Cyberspace Administration of China (CAC) jointly promulgated the “Interim Measures for the Administration of Generative AI Services,” which took effect on 15 August of that year. These are described as China’s first binding regulation to address generative AI head-on.

Its distinctive feature lies in placing under governance not only the safety of the technology but the content it produces itself. Providers offering generative AI services to the public have reportedly been required to undergo safety assessments and to file their algorithms with the authorities. It is a conception in which the state seeks to control directly the influence AI exerts on a society’s speech and order. It has been positioned as an extension of the algorithm-filing framework already in operation since 2022.

Whoever Commands the Standard Holds the Supremacy

Why do states pour such effort into rule-making? Because there is a single law that history has taught again and again: whoever commands the standard commands the market.

In the past, the standards for electricity, the protocols of telecommunications, the inspection criteria for semiconductors — the powers that seized these invisible rules went on to dominate their industries for generations. AI is no exception. How to measure safety, how to handle data, what to forbid and what to permit. The values of whoever first defines such criteria become the “arena” that later-arriving nations have no choice but to follow.

This is precisely why regulation was not a mere shackle but also a weapon of attack. Europe leveraged the size of its market, the United States its technology and alliances, and China its vast domestic market and state-led control, each striving to push its own rules up into the international standard. This tug-of-war beneath the surface is said to continue even now, through debate in diplomatic settings and international bodies.

The battle over the rules involves no force of arms and yields no flashy victories or defeats. And yet its outcome will quietly shape the form of AI — and indeed the form of society — for decades to come. Safety and speed, freedom and order: among these opposing values, the world has yet to find a shared answer.

And even as they vied over where to draw the lines of the rules, the United States and China had begun a more fundamental movement still: a force seeking to pry their technologies apart and split them into two separate worlds. Behind the trading of regulations, the United States and China had begun to pry the technology itself out of each other’s hands.