Three cheers for Attorney General James Uthmeier.
Florida’s AG, who has a bad habit of overreach, is finally in a battle worth fighting. In fact, for once, the problem isn’t that Florida has gone too far — it hasn’t gone nearly far enough.
The state’s lawsuit against OpenAI rests on a serious and credible concern: These systems can cause harm, especially to children. The complaint alleges links between AI tools and self-harm, suicide, and even violence. Investigators have pointed to cases where ChatGPT was allegedly used in planning a mass shooting at Florida State University, raising real questions about whether guardrails are strong enough — or whether they were taken seriously enough in the first place.
More broadly, the state argues that OpenAI pushed a powerful product to market while understating the risks, prioritizing growth and market share over safety.
I completely agree. Companies building these systems should be held accountable for those outcomes.
But the lawsuit, as it stands, misses two deeper issues that will determine whether this technology evolves responsibly, or simply scales its problems.
The first is authorship and intellectual property.
AI doesn’t “know” anything. It doesn’t think, reason or discover. It aggregates. It learns from vast amounts of human-created work and recombines it into something that sounds new. Those systems are trained on oceans of content — journalism, books, art, research — much of it created at real cost by people and institutions that depend on being paid for their work.
In the race to dominate this space, many AI companies have treated that content as free fuel. The more they ingest, the more powerful their systems become. Ownership, context and intent get lost along the way.
The result is an economic imbalance that should concern anyone who cares about a functioning information ecosystem. If you can take whatever you want, repackage it at scale and sell access to it, you’ve built an extraordinary business model — one that rewards aggregation over creation.
Steal everything, sell it back and call it innovation.
Meanwhile, the creators who produce the original work see their value diluted, and the financial incentive to produce high-quality journalism, research and art continues to erode.
There is a fix: Require AI companies to pay for what they take. Put a price on inputs, and behavior changes quickly. Companies become more selective. They favor credible, high-quality sources. They form partnerships instead of scraping indiscriminately.
Shoppers choose carefully. Looters grab whatever they can.
The second issue is accountability for what AI systems produce.
AI companies resist being called publishers, but functionally, that is exactly what they are. They generate and distribute information to millions of users, often in moments when those users are looking for authoritative answers.
When a newspaper publishes something false that causes harm, there are consequences. Standards apply. Legal accountability exists.
AI operates under a different set of rules.
Much of that protection ties back to a 1996 law designed for a very different internet, one that shields platforms from liability for content. Today, those protections are being tested in court as cases like this raise a fundamental question: When an AI system generates harmful or dangerous content, who is responsible?
Right now, the answer is often: no one.
That’s no longer tenable.
If AI systems can generate guidance that contributes to self-harm, reinforces destructive behavior or even plays a role — direct or indirect — in violent acts, then the companies behind those systems cannot simply point to outdated legal frameworks and walk away.
Accountability must catch up with capability.
Holding AI companies responsible for both the ingredients they use and the cake they bake won’t stop innovation. It will improve it. It will create incentives for safer design, better guardrails and more responsible deployment.
Florida is right to raise the alarm. Now it needs to aim higher.
Because this isn’t just about preventing harm; it’s about defining the rules of a new information economy. What goes in and what comes out. And whether the systems we are rapidly building will strengthen or weaken the foundation that’s already cracking under their weight.