China doesn’t need Americans to trust artificial intelligence. Its government can mandate adoption. The United States cannot. Yet some in Washington now argue that the only way to beat China is to weaken the very protections that allow Americans to trust, and therefore, use AI in the first place.

This past summer, the Senate overwhelmingly rejected an attempt to bar states from regulating AI in the One Big Beautiful Bill Act, voting 99–1 against national preemption without federal standards. Senators recognized exactly what such a move would mean: letting some of the biggest technology companies run wild. Now, Washington is debating whether to revive this idea.

A national AI preemption without federal standards gives us the worst of both worlds. It allows AI companies to escape accountability for the harms that they have caused while actively hurting U.S. AI competitiveness. We’ve already seen what this kind of hands-off policy looks like. Section 230 effectively left social media companies unaccountable for the damage their platforms inflicted. Now, some want to repeat this mistake with AI, just as public concern about AI risks is starting to reach new heights. 

If America wants to win the AI race, we need to understand what victory actually looks like. It's not just about who develops the technology first, but who gets people to use it. In China, 83% of people view AI products and services as more beneficial than harmful. In the United States, only 39% share that view. This stark contrast should concern anyone who is serious about American technological dominance. The difference is simple: China can compel adoption; America must persuade. This makes public trust a strategic necessity.

>>> What AI Means for the Future of Work

The data is clear on what drives AI adoption. Trust is one of the most fundamental prerequisites for people to adopt AI systems. When employees and consumers don't trust AI, they simply don't use it. So, how can Washington help Americans build trust in these systems? According to KPMG research, three in four people would be more willing to trust AI systems if assurance mechanisms such as adequate regulations were in place. A more recent survey shows 72% believe AI regulation is needed, with 81% saying they'd be more willing to trust AI if laws or regulations governed its responsible use.

It's clear, then, that AI preemption will not increase trust in AI and in turn, will not increase adoption of it either. This barrier to AI adoption not only threatens our ability to beat China, but it also threatens America’s economic future. Supporters of preemption argue that companies need freedom from oversight to scale fast enough to beat China. But scale without trust fails to scale at all.

The productivity gains from AI adoption are one of the main reasons AI is such a promising technology. Research shows that generative AI has increased worker productivity by 66% which is the equivalent of 47 years of natural productivity growth in the U.S. AI has also been estimated to increase GDP by 1.5% by 2035 which translates to an additional $438 billion using the current U.S. GDP figures. These figures would only increase with broader adoption of AI. However, the reality is we’re not there yet. Census data shows that only about 10% of businesses have adopted AI. This means we're capturing only a fraction of AI's economic potential. Every day that trust erodes is a day we fall further behind in realizing these gains, and AI preemption without any federal standards would accelerate that erosion, not reverse it.

There's a reasonable path forward here, but it requires doing things in the proper sequence. If policymakers genuinely believe that a patchwork of AI laws would hinder innovation, then they need to pass federal standards first, and then preempt conflicting state laws. This approach, similar to what the RISE Act does, makes logical and political sense. What makes no sense is preempting state action without comparable federal standards. This is not leadership. It’s an abdication of governing responsibility leaving Americans exposed to AI harms.

States are already responding to documented harms, like AI chatbots that encourage self-harm, generate sexual content for minors, and impersonate real people to manipulate emotions. Preempting those laws without federal replacements would leave families dangerously vulnerable.

>>> AI Companions Are Harming Your Children

AI is becoming a salient issue for voters, and most Americans want sensible safeguards for AI systems. A recent poll found that Americans oppose adding AI preemption to the NDAA by a 3-to-1 margin, with 57% opposing and only 19% supporting such a measure. Opposition to AI preemption spans all income levels and every age group, with younger Americans opposing preemption at 61%. When asked what Congress should prioritize, 90% of those surveyed chose protecting kids from AI chatbot harms over advancing and subsidizing the AI industry.

The optics of championing preemption are perilous. Telling voters that beating China requires stripping away state guardrails that affect electricity prices, job security, and child welfare without offering federal alternatives is not a winning message, and it plays directly into concerns that Washington is captured by corporate interests. 

This is not a false binary between regulation and innovation. The U.S. can do both. China isn't going to beat us because they have less AI oversight, but they will beat us if no one uses U.S. technology. Preemption without standards would hollow out America's technological future in a misguided attempt to accelerate it.

If we want Americans to adopt AI, unlock massive productivity gains, and truly win the AI race, then we need to give people a reason to trust and use these systems. This means passing common-sense federal standards before considering the preemption of state laws. It's time to stop reviving a losing argument and start building the policy framework that will secure American leadership in AI.