Logo

Democrats certainly learn from past mistakes. 

After watching President Bill Clinton let the internet grow without federal rules, Democrats want to control artificial intelligence — and if need be, to kill it in the crib.  

The only question is, where?

The executive order issued by President Joe Biden would have turned AI into a federal hand puppet serving the public policy goals of unions, educators and leftist activists.

But President Donald Trump reversed it, closing that path. 

Now Democrats are looking to blue states to regulate AI instead, balkanizing its development under a hodgepodge of state rules.

He’s aiming to preempt state authority, thus avoiding a European Union-style AI quagmire here while protecting America’s national security position vis-à-vis China. 

But Trump’s efforts are not going well.

In 1996, as the internet age dawned, Clinton declared that “the private sector should lead” its development.

Under his policy, the web “develop[ed] as a market-driven arena, not a regulated industry.”  

Clinton’s approach meshed with those of congressional Republicans: Their 1996 Telecommunications Act established federal control over interstate service, with limits on liability exposure. 

Today’s mobile communications miracles, with thousands of apps serving users on millions of phones, is a legacy to Clinton’s pragmatism.

Biden’s AI legacy would have been the opposite. 

His EO, among the longest in history, imposed a federal top-down, command-and-control regulatory approach.

It required AI models to undergo extensive “impact assessments” before going forward, and “post-deployment performance monitoring” as well. 

America’s AI would have been forced to reflect Biden’s “dedication to advancing equity and civil rights” by advancing “racial equity and support for underserved communities” and by improving “environmental and social outcomes.”  

Now, with Biden out and Trump in, the AI traps have moved to the states — with California, of course, and Colorado taking the lead. 

California has long wielded its economy — the fourth-largest in the world — in a tail-wagging-the-dog approach to push political and social change.

It’s thrown its economic weight around to make all US businesses meet California’s standards, such as its stringent CAFE auto fuel-efficiency rules.

California’s proposed AI legislation would empower its state attorney general as an AI uber-czar, making all AI models comply with Sacramento’s requirements for “safe, ethical, equitable, and sustainable” systems.   

With $1 million in fines per violation, California’s law would create massive industry uncertainty, hindering AI’s development.

Colorado’s state law involves a “Mother, may I” approach to AI innovation — a process so unworkable that its Democrat-controlled legislature and governor have agreed to delay implementation to ease its burdens on Colorado businesses.  

If AI nationwide developed under Colorado’s rules, the US Chamber of Commerce projects, the US would lose 713,000 jobs, shed $53 billion of GDP and cut national productivity —the single most important determinant of real wage growth — by 1%. 

And that’s just one state and one law, with more than 1,100 AI bills pending in state legislatures.

To avoid an EU-like jambalaya of state rules and regulations, Sen. Ted Cruz (R-Texas) tried to add a limited federal preemption of state AI laws in the One Big Beautiful Bill when it was still being hammered out. 

Under Cruz’s plan, states could still impose AI restrictions preventing unfair or deceptive practices and protecting children — provided they did not impose excessive burdens on development.  

His solution left some hot topics unaddressed — in particular, that of AI models’ use of intellectual property and their allegedly illegal lifting of copyrighted work.

But the Cruz amendment failed, and Congress remains deadlocked on a potential federal solution.

That’s why Trump is on the verge of taking action.

“There must be only One Rulebook if we are going to continue to lead in AI,” he declared Monday in a Truth Social post.

This week, the president is expected to issue an executive order to put Cruz’s proposal into practice, limiting states’ ability to handcuff AI development.

To give it teeth, the order could direct regulators to withhold federal broadband funding and other grants from states that implement restrictive AI laws. 

It’s a jerry-rigged solution at best: Federal legislation is still necessary to effectively override state AI rules.

Congress’ next shot will come in the New Year, as part of the Jan. 30 government funding bill.  

Leaving the cutting-edge technology that will determine both our nation’s future economic prospects and its military security in the hands of 50 state legislatures seems unwise and dangerous — but that appears to be Congress’ plan.

Michael Solon is a senior fellow at the Hudson Institute.

Comments
anonymous profile image
Powered by RoundtableBuilt on infrastructure designed for real-time media. Learn more at RTB.io.© Roundtable 2026. By using this site you agree to the Terms of Use and Privacy Policy