No single government has figured out how to govern artificial intelligence, and the three powers best positioned to try - the United States, China and the European Union - have each chosen a fundamentally different approach. One leans on voluntary pledges from industry, another folds AI oversight into a broader system of state control, and the third has written binding law. The divergence is not merely bureaucratic; it will shape which companies dominate the technology and which societies absorb its risks.
In the US, President Donald Trump has built AI policy around the idea that regulation is the enemy of competitiveness. On his first day back in office, he scrapped a 2023 order that required developers of the most powerful systems to share safety test results with the government. What followed was an "AI Action Plan" with more than 90 measures meant to accelerate development, alongside an order barring federal agencies from buying models deemed ideologically biased. For ordinary users trying to make sense of which tools are safe to trust, this policy vacuum matters as much as the technology itself - much as it matters when comparing privacy tools and understanding what you actually get for the price before committing to one. Congress, divided and heavily lobbied, has not passed a federal AI law, leaving a patchwork of state rules to fill the gap despite White House efforts to block them.
Security concerns reshape the American approach
The emergence of AI models capable of mounting serious cyberattacks pushed security to the center of US policy this year. A June executive order created a voluntary system allowing developers to give the government early access - up to 30 days - to their most advanced models before public release. By September, the chief executives of major AI companies had signed a White House accord promising independent audits and board-level oversight of frontier systems. The agreement carries no enforcement mechanism; Trump called it "morally binding," a phrase that captures the broader American strategy of persuasion over compulsion.
China's control-first model
Beijing has taken a markedly different path, treating AI governance as an extension of its existing machinery for managing information and online behavior. Rules on AI companion bots took effect in July, restricting emotional dependency and banning unauthorized digital clones of real people. President Xi Jinping has said AI must remain "always under human control," while Premier Li Qiang has warned of a global risk of "losing control" over frontier technologies. Yet Chinese policymakers show little appetite for slowing domestic progress, partly because many view foreign export controls as having already constrained the country's AI sector for years. State security officials have simultaneously flagged AI-generated disinformation as a national security threat.
Europe bets on binding law
The EU's AI Act, adopted in 2024, remains the world's most comprehensive legal framework for the technology, though European firms - wary of falling behind American and Chinese rivals - won a delay on rules covering high-risk systems. European Commission President Ursula von der Leyen has argued the law positions Europe to "shape global efforts," and Brussels has continued using other regulatory levers, including adding ChatGPT to its list of digital services under heightened scrutiny and proposing child-safety requirements for AI companions following a string of suicides linked to chatbot interactions worldwide. Unlike their American counterparts, European companies have generally resisted calls to slow the pace of regulation, fearing it would only entrench the dominance of US tech giants.