The White House released its AI legislative framework in late March, and the core message is clear: Congress should preempt state AI laws that burden industry. That’s the headline. But dig deeper, and there’s a much more complicated story about the tension between protecting children online and preserving free speech.

Here’s the setup. States have been racing to regulate AI in different ways. Florida proposed the strictest rules in the country—banning minors from creating AI chatbot accounts without parental consent. California’s approach is more moderate: age verification, plus required disclosures that users are talking to AI. Georgia and Arizona have their own proposals in the works.

The White House framework doesn’t reject age verification outright. In fact, it explicitly supports “privacy-protective, age-assurance requirements” for platforms likely to be accessed by minors. That’s a concession to the child safety camp.

But free speech experts are worried. The Electronic Frontier Foundation called the framework “disastrous” in a statement, specifically condemning the preemption of state protections, the age verification requirements, and a proposed federal “publicity right.” FIRE, the Foundation for Individual Rights and Expression, has been fighting age verification laws for years, calling them the “papers, please” approach—forcing adults to hand over government ID to exercise their rights online.

There’s a real human dimension to this debate. Parents testified before a Senate hearing in September about children who died by suicide after interacting with AI chatbots. A Florida mother described her 14-year-old son’s death in 2024. These stories are being used to push for stricter protections. But opponents of broad age verification laws argue that the cure could be worse than the disease—creating massive databases of people’s identities, exposing them to data breaches, and essentially building an infrastructure for online surveillance.

The core tension is this: how do you protect minors from genuinely harmful AI interactions without creating a system that tracks and verifies every adult’s identity? It’s not an easy problem. Age verification can mean uploading a driver’s license, which many adults won’t do out of privacy concerns—meaning they simply won’t use certain services. Or it can mean less invasive methods that are also less reliable.

The Trump administration is pushing hard for federal preemption. If it succeeds, states like California and Florida that have already passed different AI laws would have to conform to whatever Congress decides. That’s a massive shift in how AI regulation works in America—away from laboratory states experimenting with different approaches, toward a single federal standard.

What’s clear is that this debate isn’t going away. The deaths of those children are real. The free speech concerns are real. The industry’s desire for clarity is real. And right now, nobody has found a way to satisfy all three at once.