AI dominance: Who writes the rules?

There is a quiet but consequential shift underway in how nations think about artificial intelligence. For years, the debate was framed around risks of bias, surveillance, and misinformation.

Today, the conversation has moved. AI is now a contest over economic dominance, political values, and institutional design. The March 2026 recommendations from the White House make this shift unmistakably clear: the United States is not simply trying to regulate AI; it is trying to win with it.

Across the world, different jurisdictions are converging on very different answers to the same question: who should control intelligence in the age of machines? The US framework is built on a deceptively simple idea: govern less, build more.

Rather than imposing a comprehensive federal AI law, the approach leans on existing regulators, voluntary standards, and targeted interventions. It promotes regulatory sandboxes, expanded access to federal datasets, and incentives for private deployment. It seeks to pre-empt fragmented state-level laws in favour of a unified national standard.

The goal is to remove friction that slows innovation while intervening selectively in politically salient areas like child safety, intellectual property, fraud, and national security. Even its approach to copyright reflects this balance: rather than legislating decisively, it defers to courts to resolve whether training AI on copyrighted material constitutes fair use.

This governance treats AI primarily as an engine of growth, as reflected in the emphasis on energy infrastructure, workforce training, and small business adoption, and the insistence on protecting free speech and preventing government-driven censorship. The US is betting that the fastest innovator will ultimately shape the global rules.

If the US approach is restrained, Europe’s is defined by structure. The EU Artificial Intelligence Act represents the most comprehensive attempt anywhere to codify AI governance into binding law. It is a risk-based model. Systems deemed “unacceptable”, such as certain forms of biometric surveillance, are banned outright.

“High-risk” systems must meet strict compliance requirements. Even general-purpose AI models face obligations around transparency, copyright, and safety. Importantly, this is an institutional framework. The EU has created dedicated bodies to oversee enforcement, including a central AI Office, and has mandated regulatory sandboxes across member states to test innovation within controlled environments.

China offers a third model, combining rapid innovation with tight State oversight. It has moved aggressively to regulate algorithms, mandate registration of AI systems, and impose security and ethical requirements on developers. Unlike the US, where free speech concerns limit government intervention, or Europe, where rights-based frameworks dominate, China’s approach is explicitly State-centric. AI is seen as both an economic driver and a tool of governance. Regulations are designed not only to manage risk but to align technological development with national priorities.

This creates a paradox. China is deeply committed to scaling AI capabilities across sectors from surveillance to manufacturing. The result is a model of “controlled acceleration”: innovation is encouraged, but always within State-defined boundaries. It offers an alternative template that many developing countries may find attractive for its clarity and decisiveness, even if it raises concerns about civil liberties.

A middle path to regulation

Then, there is India, arguably the most interesting case, precisely because of what it has not yet done. India does not have a comprehensive AI law. It has adopted what might be called a “principles-first” approach. The 2025 AI Governance Guidelines under the IndiaAI Mission set out broad norms of trust, accountability, and inclusivity, but stop short of creating binding obligations. There is no standalone AI statute; governance is spread across existing laws and sectoral regulations.

This is deliberate. India’s policy architecture is built around the idea of “AI for All,” prioritising social impact in sectors like healthcare, agriculture, and education. It emphasises public-private partnerships, national data platforms, and skilling initiatives rather than strict compliance regimes.

By avoiding premature regulation, India hopes to foster innovation and attract investment. By articulating ethical principles, India signals alignment with global norms. But this middle path is not without risks. The absence of clear legal standards can create uncertainty for businesses and gaps in accountability. As AI systems become more embedded in public life, pressure will mount for more concrete rules. The question is not whether India will regulate AI, but when and on whose terms.

AI will travel where data flows, scale where incentives permit, and evolve in ways that no single jurisdiction can fully contain. The test is over whether our regulatory imagination can keep pace with a technology that has already moved beyond them.

The writer is a technology lawyer.

(Disclaimer: The views expressed above are the author's own. They do not necessarily reflect the views of DH.)