The AI Deregulation Scam

Why Trump’s executive order is a giveaway to tech billionaires, not a win for innovation

You’ve probably noticed that I’ve been pretty bullish on AI. I use it for research, as an editor, and for all of the incredible images that people heap praise on me for (just kidding!). But I do think the technology is genuinely useful, and I’m not someone who reflexively panics about things.

However, with that being said, what Trump announced on Monday is a genuinely bad deal for everyone except the companies that wrote it.

The president confirmed he’ll sign an executive order this week to create “ONE RULE” for artificial intelligence that would block states from regulating the technology. “You can’t expect a company to get 50 Approvals every time they want to do something,” he posted on Truth Social. The draft order directs Attorney General Pam Bondi to create an “AI Litigation Task Force” whose job is to sue states over their AI laws. States that don’t fall in line could lose some of their federal broadband funding.

Here’s what makes this particularly galling. This isn’t some Trump idea. It’s a copy-paste from tech industry lobbyists, most aggressively from Andreessen Horowitz and OpenAI, who’ve been pushing this exact framework for months. They submitted it to the White House AI Action Plan. Andreessen Horowitz and OpenAI’s president, Greg Brockman, helped put more than $100 million into a super PAC network called Leading the Future to back candidates who oppose AI regulation. They got what they asked for, dressed up as executive policy.

The whole premise is a scam. Tech companies claim they need federal preemption because a “patchwork” of 50 state laws would cripple innovation and let China win the AI race. What they actually want is to replace those 50 sets of laws with zero regulation, because the federal government has passed next to nothing. There is no federal AI framework. There’s no broad AI consumer protection law at the national level. The executive order would eliminate state safeguards and replace them with a promise that Congress might do something eventually.

Meanwhile, those state laws actually protect people.

Colorado passed the first comprehensive AI anti-discrimination law, requiring companies to protect consumers from algorithmic bias in hiring, housing, and healthcare.

California is cracking down on AI companion chatbots that encourage self-harm in children. Texas banned AI systems designed to produce deepfake child pornography. Tennessee protects musicians from having their voices cloned without consent.

Forty-five states have criminalized AI-generated child sexual abuse material, according to the group Enough Abuse.

These are the “burdensome” regulations the tech industry wants eliminated. Laws against deepfake child porn. Laws preventing AI from discriminating against job applicants. Laws that say chatbots can’t groom teenagers.

The legal argument is also garbage. A president cannot preempt state laws through executive order. Full stop. That’s Congress’s job. And even Congress has limits – in Murphy v. NCAA, the Supreme Court held that Congress can’t just order state legislatures not to pass laws. An executive order can direct federal agencies to take certain actions, but it cannot override state statutes. The order will face immediate legal challenges and will almost certainly lose.

Congress already rejected this approach. In July, the Senate voted 99 to 1 to strip a 10-year AI moratorium from Trump’s “One Big Beautiful Bill.” That’s not a typo. Ninety-nine senators, including Republican Marsha Blackburn and Democrat Maria Cantwell working together, said no. Ted Cruz, who championed the moratorium, couldn’t even get his own colleagues to support it. The only senator who voted to keep it may have done so by mistake, a source told TIME – he voted after 4 a.m. during an all-night session.

Dur….

Ron DeSantis, not exactly a progressive warrior, called the effort “federal government overreach” and proposed his own Florida AI Bill of Rights. “Stripping states of jurisdiction to regulate AI is a subsidy to Big Tech,” he said. Marjorie Taylor Greene, of all people, posted that “Federalism must be preserved.” These aren’t Democrats carrying water for regulation. These are Republicans who recognize that eliminating state authority to protect citizens from corporate harm is a bad idea regardless of who benefits.

The AI bubble I’ve been warning about may also be about to pop. SoftBank sold off its entire Nvidia stake. An MIT report found that 95 percent of generative AI pilot programs aren’t generating returns for companies. OpenAI executives recently floated the idea that the government should “guarantee” their AI infrastructure investments, which everyone understood as a request for a bailout. So the timing of this executive order, conveniently shield the industry from accountability right before the hype cycle collapses, is probably not a coincidence.

I want AI to succeed. I think it has enormous potential. But potential is not a blank check. These companies are asking to be exempt from laws that prevent them from hurting children, discriminating against workers, and manipulating vulnerable people. They’re doing it by pretending that basic consumer protections are somehow anti-innovation. And they’ve convinced the president to do their dirty work through an executive action that’s constitutionally questionable at best.

Being pro-technology and pro-accountability are not mutually exclusive positions. The tech industry has just spent years convincing people otherwise.

Sources

Correction, September 27th, 2026: This essay first said Andreessen Horowitz and OpenAI funded a $100 million super PAC – the money came from Andreessen Horowitz, OpenAI’s president Greg Brockman personally, and others. It also called Maria Cantwell a Republican – she’s a Democrat – and said Trump announced his order “yesterday,” when he posted it the same day this essay went up. And it said the Supreme Court’s Murphy v. NCAA ruling held that the federal government can’t stop states from legislating – the case held that Congress can’t give direct orders to state legislatures, and it doesn’t address presidential orders. It also said 38 states had criminalized AI-generated child sexual abuse material – by the time this essay went up, the count was 45.

Update, September 27th, 2026: A few lines now say who says so and note the one federal AI deepfake law Congress has passed. Two more now say how much broadband money was at stake and who sold Nvidia stock.


Everything I have written about AI is in one place at rachelandthecity.com/the-ai-files.