A Federalism Fight Splits MAGA...

And confuses Kaul

I hope everyone reading this had a truly great Thanksgiving with those you love, if not in their physical presence, then at least in the warm embrace of shared thoughts.

I myself had a great holiday with the family: Turkey, stuffing, mashed potatoes, and, for dessert, a third or maybe fourth helping of Wisconsin attorney general Josh Kaul.

I know, it seems like I write about him a lot. I try not to, but the poor fellow just keeps offering up more half-baked goodies, delivering without fail like a Dessert of the Month Club selection. This month he glided into another hot oven—state regulation of artificial intelligence (AI)—and, for some unfathomable reason, signed a letter with many of his top-cop counterparts that actually contained some reasonable arguments.

Obviously, he is a confused man.

He would likely be horrified to know that more than a few Republican attorneys general also signed the November 25 letter to congressional leadership urging Congress not to preempt the state regulation of artificial intelligence with a moratorium, which some of said leadership and President Donald Trump seem determined to do. Among the Republicans Kaul is hanging with in this letter are Todd Rokita of Indiana, Liz Murrill of Louisiana, Lynn Fitch of Mississippi, Alan Wilson of South Carolina, Dave Yost of Ohio, and Jonathan Skrmetti of Tennessee.

I’ll get back to the attorneys general letter in a sec, but first, the whole issue of state AI regulation is not getting the attention it deserves. The synopsis here is that Big Tech, Trump, and a group of congressional lawmakers are trying to prevent state regulation of AI in favor of a standardized federal framework. However, the populist left and the populist right have pushed back, albeit for largely different reasons, and the outcome is uncertain.

The backdrop to the story is that AI is racing ahead as potentially the most disruptive force of this era—for better or for worse—but, as it does, the new technology is running headlong into the oldest of constitutional debates: who regulates what. Beyond debates about algorithms and the computing power of large language models (LLM) is the ongoing tension between states’ rights and the federal government, that is to say, the very boundaries of our federalism.

The fight has been simmering for a while, but it erupted in full volcanic fury during this past summer’s debate over Trump’s One Big Beautiful Bill, which contained a provision banning state artificial intelligence laws for a period of up to 10 years. That push was led by Trump, Sen. Ted Cruz (R-Texas), and leading technology companies, who argue that a patchwork of state rules threatens America’s AI advantage over China.

The provision was jettisoned 99-1, with the leadership of Republicans like Tennessee Sen. Marsha Blackburn and, on the state level, Arkansas Gov. Sarah Huckabee Sanders’s and Florida Gov. Ron DeSantis’s passionate opposition. But it never went away, and now it’s back again, both in the guise of a potential Trump executive order and as a potential provision in the National Defense Authorization Act (NDAA).

Of course, it also hasn’t taken long for a wall of resistance to again form, which includes Republicans, Democrats, libertarians, left-wing and right-wing populists, anti-Big Tech conservatives, state legislators across all 50 states, and now, in a cosmic plot twist, our very own Mr. Kaul, who has finally taken a position that doesn’t inspire half the state to pay his way to a Comic Con, preferably in a land far, far away.

“Prohibiting states from putting in place laws that can help protect against dangers associated with AI would be a major mistake,” Kaul warned this past week. “Congress shouldn’t be sacrificing the interests of the public as a whole in order to benefit big tech.”

That puts Kaul and many MAGA Republicans on the same side of the issue, and, when that happens, it’s a sure thing that something strange is brewing, and it’s not your morning coffee. The bottom line is that the federal preemption fight has become the newest, largest, and perhaps most consequential test of American federalism in decades.

And, as usual, the Democrats aren’t being honest.

A Draft Executive Order

The immediate and most recent firestorm began when reporters uncovered last week a draft executive order circulating within the administration, tentatively titled “Eliminating State Law Obstruction of National AI Policy.” If nothing else, that’s very anti-states’-rights-sounding.

The order would have instructed the Justice Department to establish a task force to identify and challenge state laws that conflict with a new uniform national policy framework, including laws that “unconstitutionally regulate interstate commerce.” The administration would also identify state laws that require AI models to alter their truthful outputs, or that “may compel AI developers or deployers to disclose or report information in a manner that would violate the First Amendment….” The order would also potentially withhold federal technology funds, including broadband grants, to pressure states into retreat.

Reuters reported that the draft showed “how far Trump was willing to go” to help AI companies overcome a patchwork of laws they say stifle innovation, and noted that the order would have faced “significant pushback from states.”

That much is true, but it’s also fair to point out that the executive order also took aim at states that might try to bake DEI and woke ideology into the AI cake and to manipulate objective truth in doing so. Here’s how the draft executive order put it:

“And a new Colorado law about so-called ‘algorithmic discrimination’ may even force AI models to embed DEI in their programming, and to produce false results in order to avoid a ‘differential treatment or impact’ on Colorado’s enumerated demographic groups. But the United States’ Al regulatory framework must prioritize truth.”

That offers up the real reason why Democratic attorneys general like Kaul issued the November letter—they likely don’t want federal action that might disrupt state-level initiatives to implant leftist doctrine into AI models, particularly in blue or swing states. To say it another way, GOP opposition is being motivated by states’ rights principles and real harms, whereas Democratic opposition is more likely motivated by the convenience of supporting states rights in this one case for a potentially nefarious purpose—ideological infiltration that causes some of the very harms the attorneys general point to in their letter.

But that doesn’t resolve the problem at hand. No matter its nobleness in seeking objectively truthful AI models, there’s a states rights problem in the proposed EO and in fact in any kind of federal preemption, and a lot of conservatives aren’t willing to look the other way.

Reuters was right about one thing. Once the leak hit the press, the pushback was hard, and Trump himself stepped into the arena. The draft order was shelved, at least for the time being, but the president used his megaphone on Truth Social to press Congress to include preemption in the pending NDAA. He wrote:

“Investment in AI is helping to make the U.S. Economy the ‘HOTTEST’ in the World—But overregulation by the States is threatening to undermine this Growth Engine… Put it in the NDAA, or pass a separate Bill, and nobody will ever be able to compete with America.”

House leadership, particularly majority leader Steve Scalise (R-Louisiana), had already been searching for a vehicle to carry the preemption language, according to Punchbowl News. After previously floating the reconciliation bill and other avenues, Republican leaders began openly discussing folding it into the NDAA.

Of patchworks, China, and “Crushing Innovation”

Supporters of federal preemption, including conservatives such as Sen. Ted Cruz (R-Texas) and Big Tech companies, have built their case primarily around a potent geopolitical rationale, namely, a perceived AI race with China.

First is the patchwork argument. If every state creates its own rules for AI transparency, algorithmic auditing, data privacy, child protection, and scamming restrictions, so the argument goes, companies face 50 different compliance regimes. The fear is that AI firms, especially startups, would be buried under a regulatory obstacle course so massive it would be impossible to navigate. That’s bad for business as well as for innovation.

Cruz, chairman of the Senate Commerce Committee, has argued that the Commerce Clause gives Congress the authority—and perhaps the duty—to impose a single standard. As he put it during the battle over the provision in the reconciliation bill:

“The Constitution gives Congress the authority to regulate commerce between the states, and AI is quintessentially commerce between the states,” Cruz said in June, as reported by The Hill. “ … Having a patchwork of 50 different standards would be devastating to the development of AI.”

To Cruz, AI—like aviation, interstate telecommunications, and railroads before it—demands national uniformity. Tech firms echo that view. From Google to OpenAI, companies have publicly supported a federal framework, saying it would reduce compliance costs and eliminate the administrative chaos.

Critics acknowledge that that is a powerful argument, but they say it is also incomplete. That’s because, in this view, the harms AI produces, everything from election disinformation, scam texts and calls, deceptive consumer product information, the compromise of data privacy, child-directed content, algorithmic discrimination in housing, and more, play out in states, cities, neighborhoods, homes, and not in the Washington, D.C. swamp.

Florida Gov. Ron DeSantis has been one of the most aggressive conservatives opposing pre-emption, arguing it would muzzle states trying to protect their residents. DeSantis warned on X:

Though she has lost a lot of her footing in MAGA recently, along the way earning the president’s wrath, soon-to-be-former U.S. Rep. Marjorie Taylor Greene (R-Georgia) has been fighting pre-emption from the get-go. She posted on X:

“There should not be a moratorium on states rights for AI. States must retain the right to regulate and make laws on AI and anything else for the benefit of their state. Federalism must be preserved.”

So too has Steve Bannon, who, on his podcast, said: “They’re trying to slip this into another 9,000-page, must-pass NDAA, that they don’t want to have a discussion about, in the dark of night.”

Bannon warned that the move was an attempt to shield AI giants from public scrutiny.

Wisconsin U.S. Sen. Ron Johnson has opposed preemption, too. Though he has taken a more procedural approach, he nonetheless lands in the same place, opposing a federal ban on state AI laws. In an interview with Wisconsin Public Radio, Johnson criticized the 10-year moratorium buried in Trump’s budget bill: “I don’t like it because it hasn’t been discussed. It hasn’t been debated. This is an incredibly complex issue, AI.”

He doubted the measure would survive Senate budget rules because its fiscal impact was unclear, a polite way of saying it was jammed into the wrong bill. But then came the constitutional pivot. Speaking to The Hill, Johnson said:

“I personally don’t think we should be setting a federal standard right now and prohibiting the states from doing what we should be doing in a federated republic. Let the states experiment.”

In a June 3 letter signed by more than 260 state legislators from all 50 states, including Wisconsin Republican lawmakers Dan Knodl and Shannon Zimmerman and Democrat Kelda Roys, lawmakers warned Congress that the proposed 10-year freeze of state and local regulation of AI would cut short democratic discussion of AI policy with a sweeping moratorium that would restrict policymakers from responding to emerging issues.

“As state lawmakers and policymakers, we regularly hear from constituents about the rise of online harms and the impacts of AI on our communities,” the lawmakers wrote.

The federal preemption provision—which at the time was proposed in the budget reconciliation bill—would overturn laws already on the books, the lawmakers wrote, including statutes that increase consumer transparency, set rules for government acquisition of new technology, protect patients in the healthcare system, and defend artists and creators. Valuably, the lawmakers framed the age-old constitutional question properly:

“States are laboratories of democracy accountable to their citizens and must maintain the flexibility to respond to new digital concerns,” they wrote.

In short, let the states work.

All of which brings us back to Josh Kaul. The Wisconsin attorney general and his colleagues in their November 25 letter to congressional leadership warned that a moratorium on state laws would have severe consequences.

“We are also deeply troubled by sycophantic and delusional generative AI outputs plunging individuals into spirals of mental illness, suicide, self-harm, and violence,” they wrote. “And we are concerned that AI chatbots and ‘companions’ are engaging children in highly inappropriate ways, including with conversations that feature graphic romantic and sexual role play, encouragement of suicide, promotion of eating disorders, and suggestions to prioritize use of the AI at the expense of connecting with friends and loved ones in real life.”

Reading this part of the letter convinced me that Kaul was confused or delusional. He opposes AI “engaging children in highly inappropriate ways, including with conversations that feature graphic romantic and sexual roleplay,” but apparently doesn’t think it’s inappropriate when government educators insert gender identity ideology into the classroom or when the state endorses the physical mutilation of minors under the rubric of gender-affirming care.

But, by God, by all means let us regulate AI to protect the children. One wonders whether he is really concerned about what AI is doing or just worried about who is controlling it. Indeed, the potential control of AI by people such as Kaul are one of the reasons the administration wants preemption.

Despite Democratic hypocrisy, the November 25 letter makes valid points that supporters of pre-emption must overcome. Indeed, Big Tech’s arguments are similar to those we have heard from Big Pharma and Big Food to kill state consumer protections and curry federal favor from bureaucratic collectivist allies in Washington.

For example, recently Merck, Pfizer, and Roche, complained to the Department of Justice about a “patchwork” of state PFAS laws, the most restrictive of which they argued imposed a nationwide PFAS restriction because companies “must often adjust operations to comply with the most restrictive state law.”

Federal law should override state mandates, the conglomerates argued. It’s the same argument Big Tech is using right now.

That’s always the mantra of why state sovereignty should be dismantled once and for all, and the 50 states should be essentially be relieved to serve as colonies of Washington. There have been voices of reason on this issue in Congress, Blackburn, Missouri’s Josh Hawley, Kentucky’s Rand Paul, and our own Ron Johnson among them.

A win-win compromise

Along with DeSantis and others, they are holding firm on states’ rights, as they should. However, there just may be a way to compromise that satisfies everyone and protects states’ rights.

As it turns out, Cruz had the answer all along—the federal government regulates interstate commerce. That’s not the problem. The problem is that Big Tech wants the federal government to regulate not only issues relating to interstate commerce—the development of large language models, for instance—but the use and misuse of already developed technology within state borders.

Writing this week, Jennifer Huddleston, a senior fellow in technology policy at the Cato Institute, made astute observations that could point the way to cleaving federal jurisdiction from state jurisdiction and to preserving the constitutional prerogatives of both. Basically, Huddleston observed that while states increasingly seek to meddle in federal-jurisdictional interstate concerns, many of the real state concerns are intrastate.

“Many opponents to preemption or a federal moratorium on AI policy point to concerns about specific harms, such as fraud, discriminatory use, or specific harmful applications,” Huddleston wrote. “These opponents also bemoan the ‘failure’ of federal action on AI and see the states as filling a necessary void.”

In other words, Huddleston wrote, issues such as computer power, model development, and other key components of AI innovations cross state borders in the interactions needed both to develop and deploy the technology.

“State laws regulating model development, computing power, and other general aspects of AI would impact AI technologies well beyond their borders in ways that raise serious interstate commerce concerns,” she wrote. “If states are to provide positive models of what AI governance could look like or address specific concerns, this policy approach would need to be limited to intrastate issues.”

She pointed to safeguards for civil liberties around state government use of AI, the use of state government-collected data, or updating state laws to allow for AI or clarify that AI does not absolve a bad actor of their violation. Not to mention that many of the abuses surrounding scamming or luring children are already illegal, no matter what technology one is using.

Huddletson’s viewpoint and mine might not totally align, but states could enhance penalties for AI deployment for such purposes within their borders, but the bottom line is that most of those questions are indeed intrastate. They don’t have anything to do with innovation and computer and model development; they have everything to do with the use of those models and computers after they are developed.

That’s the critical distinction.

There are laws already on the books that address intrastate concerns that preemption could wipe away if the distinction is not observed. Utah passed the first AI law protecting children’s data. Tennessee and Georgia passed deepfake and synthetic-media laws, which in Tennessee protects artists’ voices from AI cloning, and Georgia makes it a felony to distribute intentionally deceptive AI content before an election. Florida faces AI-enabled election misinformation. North Carolina faces voice-clone scams targeting seniors.

One set of federal rules will not address all these problems, nor their particularities in the individual states. What’s more, AI innovation has not been hurt by any of these statutes.

A question always to ask in considering any state law: Is AI a source of harm to consumers, children, renters, voters, and patients that fall within the states’ traditional police powers? The truth is, the Founders assumed that states would handle health, safety, welfare, and morals.

So really, once we kick out Big Tech’s Big Fake that all the issues surrounding AI regulation are interstate, the fix is simple. Instead of a broad all-encompassing federal preemption, just pass federal laws—the federal framework the industry and the administration want so badly—that regulate AI development along clear lines delineating interstate and intrastate jurisdictions, in other words, a national standard that doesn’t throw out the baby with the bath water.

And, if we are truly to be aligned with our federalism, if a state insists on passing a conflicting statute, the onus is on the federal government to prove it, and then preempt that single statute in a court of law. All of that allows Big Tech to develop and compete. It protects state sovereignty, and it keeps our unique federalism intact.

Not least, it would also help keep Josh Kaul’s brain from exploding when he realizes he’s on the same side of an issue as the far right.

Sometimes the truth is stranger than fiction. But, as they say, the truth is out there. In the end, the question is the same one that has hovered over this country since the beginning: Who decides?

Most of the time, it must be the states. All of the time, when the special-interest smoke clears away, the decision is a simple one.