Washington's inaction on artificial intelligence regulation stands in stark contrast to the American public's clear preference for safety over global competition. A recent poll found that 61% of Americans would rather be shielded from AI's dangers than see the U.S. outpace China, with even a plurality of Trump voters—44% to 31%—agreeing. This sentiment was echoed in the midterm elections, where voters in at least 38 localities approved bans or restrictions on data centers, and 56% said AI and data centers were a major or crucial factor in their vote.
Despite this, tech companies continue to rush more powerful models to market, often with little oversight. OpenAI CEO Sam Altman has suggested that "the world should accept some bad things happening" in exchange for AI's benefits, a stance that critics say prioritizes corporate profits over public safety.
Why has Congress failed to act? The answer lies in the immense financial and political power of tech giants, a stock market buoyed by AI hype, and an administration cozy with Silicon Valley. But opponents of regulation often hide behind the specter of China. President Trump has declared that "whoever wins that race wins," while Anthropic CEO Dario Amodei warns of "geopolitical dominance" if China pulls ahead. Sen. Ted Cruz (R-Texas) has been even more blunt: "If there are gonna be killer robots, I'd rather they be American killer robots than Chinese killer robots."
This framing, however, rests on several flawed assumptions. First, it presumes that whoever reaches advanced AI first achieves some decisive, undefined victory. But as a Foreign Affairs analysis notes, China already possesses formidable cyber, military, and economic tools, yet has largely refrained from using them to cripple U.S. interests, knowing a devastating response would follow. U.S. deterrence would not vanish simply because China developed a new technology.
Second, the race assumes American AI will be safer than Chinese AI. But the companies themselves admit they don't fully understand or control what they're building. A former Anthropic researcher recently stated that those building AI "earnestly believe that it could kill us all by the end of the decade." Technology that we don't control is not loyal to any nation—it poses risks to everyone.
Third, the two AI ecosystems are deeply intertwined. U.S. export controls limit China's access to advanced chips, but Chinese firms have also trained their models on the outputs of American ones and rely on U.S. cloud providers. Tougher restrictions could slow China's progress, as seen in the ongoing debates over Nvidia chip sales to China.
Ultimately, Americans want guardrails, even at the cost of ceding ground to Beijing. Beating China ranks among the least important AI concerns for voters. Because the world's leading AI companies are American, Washington can regulate much of the frontier on its own. Experts like Michael Fuchs, director of U.S. Foreign Policy at the Open Society Foundations and a former Biden aide, argue for a pause on frontier development until robust safeguards are in place.
Such a pause would address not only existential risks but also more immediate harms: massive job losses, threats to civil liberties, the strain of data centers on local communities, labor exploitation, and the outsized power these companies wield over our economy and daily lives. New laws and enforcement of existing ones are needed to hold companies accountable for the harms their products cause.
China, meanwhile, is not ignoring AI's dangers. It regulates AI domestically and has imposed its own export controls, fearing threats to political stability. A serious U.S.-China agreement on AI safety may be distant, but an American pause—taken in America's own interest—could signal credibility to Beijing. History offers a precedent: the U.S. unilateral renunciation of biological weapons in 1969 helped pave the way for the Biological Weapons Convention.
We don't fully understand this technology, but we still decide how it is built and deployed. Fear of China should not drive us to build something we know is dangerous. If America can lead in AI innovation, it can also lead in AI safety—a priority the public clearly supports.
