Majority Leader Steve Scalise is leading the charge to shoehorn preemption of state regulation of artificial intelligence into the National Defense Authorization Act (NDAA), a must-pass bill. The text of the preemption measure is not public, so American voters simply do not know what exactly is being haggled over on Capitol Hill.

This fight will determine who has a voice in AI policy in the years ahead, as this new technology reshapes our lives in untold ways.

President Trump’s voters have an instinctual understanding of the risks that come with getting AI policy wrong. Many polls have found that Americans fear that AI will cause major job loss, harm our children, and lead to a less hopeful future. The Senate already resoundingly rejected a similar preemption provision in June with a sweeping vote of 99-1. Republican state lawmakers, attorneys general, and governors, likewise, do not want to have their hands tied on an issue that is steadily growing in importance.

At the Institute for Family Studies, we have polled preemption with YouGov on three separate occasions, and each time found overwhelming bipartisan opposition to it. The latest poll, released this past Friday, specifically asked voters about including preemption in the NDAA. Unsurprisingly, we found overwhelming opposition to the measure, including among Trump supporters, 43% of whom oppose including preemption in the NDAA, while a mere 25% support it.

It’s clear that this is an issue where Republican leaders in Washington are deeply disconnected from Republican voters. But how to explain their determination to ram through a top-down AI policy by hook or by crook? The only way to make sense of it is to consider that the AI industry driving these policies does not believe it can get what it wants if the legislation is subjected to the normal democratic process.

In Washington, NDAA negotiations are usually the ultimate “process story,” characterized by professional bipartisanship between House and Senate negotiators of both parties. But this time, at the 11th hour, it has been turned into a sideshow -- a proxy fight over who gets to decide the future. This behind-closed-doors effort to form AI policy unilaterally from the top down feels intentionally antidemocratic. It is a huge risk for a party that won the White House and slim majorities in both the Senate and House on the back of populist sentiment.

Republican leadership, despite that, appears poised to try to block Republican governors and representatives and red states from regulating AI, while also blocking the American people from having a view into what will be the most consequential legislation of this generation.

Where one stands in the AI preemption battle depends on what one wants to protect. Should we prioritize protecting the AI industry from state lawmakers, or should we prioritize protecting kids, workers, and American citizens from unregulated AI systems?

And yet, both sides ultimately want a federal AI standard. Industry wants a uniform regulatory environment. Those who oppose preemption believe that the states have a vital and enduring role, but they are willing to trade certain state-level authorities in return for even stronger federal regulations arrived at through open debate and citizen participation. 

But we also disagree on whose voices matter in the formation of federal AI policy. Recent actions by the pro-preemption camp (who style themselves as “accelerationists”) suggest they think that every aspect of federal AI legislation should be decided in a backroom deal between the White House, congressional leadership, and industry lobbyists. Preemption opponents, by contrast, believe that the regular legislative process is the proper vehicle for the establishment of a federal AI standard, because whatever form it takes will affect the lives of children, families, workers, and communities nationwide.

Everyday people don’t have congressional leaders on speed-dial (or an army of lobbyists to press their cause). The normal legislative process is the only chance for them to have a voice in regulations that will affect their jobs, families, and lives forever.

Fortunately, President Trump thinks establishing a federal standard should be a priority for Republicans -- but he understands that it need not be accomplished through the NDAA. In a recent post on Truth Social, the president clearly stated that he wanted Congress to prioritize establishing a federal standard for AI regulation, but that he was ambivalent about whether the best way to do it was through the NDAA or to “pass a separate bill.” But by all appearances, despite President Trump’s openness to the normal legislative process, Republican leadership shows no interest in it.

There is a better way. Leaders of the Republican Party should not preempt state AI laws until a federal standard is in place, and they should trust the democratic process on an issue that will have foundational implications for the lives of every single one of their voters -- and everyone else.

Figuring out how best to regulate AI requires more democracy, not less. President Trump has said that he is open to passing a separate bill that follows the regular rules of order. For the sake of this country and our shared future, we must take this better path.

Michael Toscano is senior fellow and director of the Family First Technology Initiative at the Institute for Family Studies.

Image: Public Domain Pictures