The House GOP Quietly Slipped In An AI Law That Would Accidentally Ban GOP’s Favorite ‘Save The Children’ Laws

Policy

from the did-they-have-AI-write-this-too? dept

Buried in the House GOP’s massive budget reconciliation bill is a seemingly simple provision about AI regulation. The idea appears straightforward enough: stop states from regulating AI companies for the next decade. To do this, they quietly added language preventing states from regulating “artificial intelligence models, artificial intelligence systems, or automated decision systems.”

This is, ostensibly, about “protecting innovation.” Or maybe just pleasing some campaign donors [waves to Elon!]. But there’s a small problem.

What, exactly, is an “automated decision system”?

According to the bill, it’s “any computational process derived from machine learning, statistical modeling, data analytics, or artificial intelligence that issues a simplified output, including a score, classification, or recommendation, to materially influence or replace human decision making.”

Right. So, yes, that would include AI.

But you know what else it includes? Pretty much every content moderation system ever created. And a whole lot more as well. As Dreamwidth founder Rahaeli points out, the incredibly broad and vague language would actually ban enforcement of basically any state internet regulation for the next decade. It could ban laws around using technology to diagnose health. Or using AI for hiring or sentencing guidelines.

Amusingly, the state laws this would kill include Republicans’ favorite new (generally unconstitutional) hobby horse — all those “protect the children online” bills they keep passing. You know the ones: laws requiring social media companies to verify ages, scan for harmful content, and generally “think of the children.”

See, every single tool these laws require would count as an “automated decision system.” The algorithm that checks if someone’s underage? That’s an automated decision system. The tool that flags potentially harmful content for kids? Also an automated decision system. The filters that GOP state legislators demand social media companies use? You guessed it.

In fact, if you read the definition carefully, it would seem to ban state regulation of pretty much any computer system that helps make decisions. Which, in 2025, is… most computer systems?

At this point, you might be thinking, “Well, surely there are exemptions for…” but nope. While the bill does include some exemptions, none of them would save these state-level content moderation laws. The House GOP, in their rush to please AI companies, have written a law so broad it would effectively prevent states from regulating any algorithmic system for a decade.

There’s a certain poetic justice here. These state-level social media laws have consistently been struck down as unconstitutional anyway. Having them all preempted by federal law might actually save everyone some time and legal fees.

At the very least, there’s a temptation to sit back and “let them cook.”

But this creates a fascinating contradiction: Republican state legislators are pushing for more control over tech companies, while Republican federal legislators are accidentally making that control impossible. It’s almost as if they don’t actually have a coherent technology policy beyond “do whatever seems politically expedient at the moment.”

To be fair, having federal-level AI regulation rather than 50 different state laws probably does make sense. But there’s a difference between thoughtful federal preemption and… whatever the fuck this is.

The House GOP clearly wanted to score some quick points with the AI industry. Instead, they’ve written a law that would nuke their own party’s cherished “save the children” crusade. Though given how those laws keep getting struck down anyway, maybe they’ve accidentally done everyone a favor.

Filed Under: ai, budget reconciliation, content moderation, gop, house gop, internet regulations, moratorium, protect the children, state laws