The AI Bet With China: Why America Needs Rules Before We Talk

What Happened

This week, U.S. and Chinese officials are meeting in Washington to discuss what the administration is calling "more transparency" on artificial intelligence. The proposal includes an "AI incident alert system", a mechanism for each country to notify the other when something goes wrong with their AI systems, presumably to avoid miscalculation or escalation.

On the surface, this looks reasonable. Two nuclear powers talking about AI safety. What could go wrong?

The answer: everything, if America walks into that room without knowing what it actually stands for on AI.

What It Means for You

Here's what's at stake for an ordinary American: the rules we live under for AI will be written this decade. Right now, they're being written by six companies, OpenAI, Google, Meta, Microsoft, Anthropic, and xAI control 90% of frontier AI development. Not by elected government. Not by you.

The European Union already passed its AI Act in 2024. China is building its own regulatory framework. And the United States? We have no federal AI law. None. Thirty states are writing their own rules, which means you live in one patchwork and a company lives in another, and guess who wins that negotiation.

Meanwhile, 63% of Americans say AI needs more regulation. That's not a fringe position. That's a supermajority of the country, right and left, telling Washington: figure this out before it figures us out.

The people this hits hardest are workers. Thirty percent of hours worked could be automated by 2030 according to White House economic analysis. But only 24% of displaced workers receive any retraining. We're sleepwalking into a job displacement that dwarfs anything since factory automation, with almost no safety net in place.

The Bigger Picture

Talking to China about AI safety is actually smart policy. China is both a competitor and a necessary partner, there's no way around that. The false choice between Cold War confrontation and naive accommodation serves nobody. If two countries with the most advanced AI capabilities can't communicate about what happens when something goes catastrophically wrong, the rest of us pay the price.

But here's the diplomatic trap: you can't negotiate from a position of strength about rules you haven't written for yourself.

When the EU sits down with China or America on AI, they're backed by the EU AI Act. They have legislation. They have enforcement mechanisms. They have a legal framework that says: this is what we require, this is what we prohibit, and this is how we check. When America sits down, we're backed by... hope? And the voluntary commitments of six companies who have every financial incentive to keep regulation light?

Denmark spends 2% of its GDP on active labor market policies, job training, wage support, redeployment programs, to handle workers displaced by automation. The United States spends 0.1%. That's a 20-fold difference. We're about to automate 30% of the work done in this country and we're spending the equivalent of a rounding error on preparing people for it.

This isn't just about fairness. It's about whether we can actually hold China or anyone else accountable for what their AI does if we don't have clear rules about what our own AI can do.

Where This Goes

The meeting this week matters. Transparency with China on AI incidents is better than opacity. But it's a band-aid on a fracture.

What America needs before it negotiates with anyone is a federal AI law. Not a law written by six companies and rubber-stamped by Congress. A real law that says:

High-risk AI systems, ones that touch hiring, lending, medical decisions, criminal justice, hiring, require transparency about how they work and regular audits to make sure they work fairly. Companies disclose when they're training AI systems that displace significant numbers of workers, and fund the retraining. We fund a jobs transition program that matches what Denmark spends, so workers aren't left holding the bag. We actually enforce it.

That's not radical. That's what the EU did. That's what 63% of Americans are asking for.

Until we do that, every conversation with China is us negotiating without leverage. We're hoping they'll be transparent about their AI while ours operates in the dark. We're asking them to retrain their workers while we leave ours behind. That's not a negotiating position. That's a confession.

The diplomacy this week is real and important. But the real work, the hard, boring, legislative work of actually writing rules, that's what happens in Congress. And that's where this is really decided.