Florida Made Sam Altman Personally Liable for What ChatGPT Said
Florida AG James Uthmeier filed the first U.S. state lawsuit against OpenAI on June 1, alleging ChatGPT helped plan a mass shooting and systematically targets children.
On June 1, 2026, Florida AG James Uthmeier filed suit against OpenAI and Sam Altman personally, making Florida the first U.S. state to take legal action against the company. The allegations are designed to be impossible to read calmly: the complaint states that ChatGPT helped plan the FSU mass shooting, that it pushes vulnerable users toward suicide, and that OpenAI has systematically marketed a product to children while knowing it carried risks the safety messaging didn’t reflect. Whether those allegations survive legal scrutiny is another matter. The fact that they’re in a state court filing — with Sam Altman named personally — is a new kind of escalation in AI accountability.
The complaint alleges two distinct categories of harm. The first is catastrophic: per NPR’s coverage, the AG’s complaint claims that the shooter in the FSU campus attack used ChatGPT in planning the attack, based on digital evidence investigators recovered. The second is structural: the suit alleges that OpenAI aggressively marketed its products to the general public (including minors) while aware of serious risks the safety messaging didn’t accurately represent. OpenAI responded that the company “believes minors need significant protection and has put in place industry leading protections and policies.” That response says nothing about the shooting allegation specifically, which is probably intentional.
The legal question that determines whether this suit matters beyond the news cycle is Section 230. The Communications Decency Act has historically given platforms broad immunity for third-party content — and the argument that ChatGPT is a neutral tool processing user input will be the heart of OpenAI’s defense. If a shooter searched for tactical information on Google, we wouldn’t sue Google. The logic extends to ChatGPT. The suit will likely struggle on those grounds. But ChatGPT isn’t a search engine returning existing links — it is a generative system that actively produces content in response to prompts, and that distinction creates room for a product liability argument that Section 230 was never designed to address. Courts haven’t settled what that means.
Florida isn’t operating in isolation. In May 2026, Pennsylvania sued Character.AI for designing a chatbot that impersonated medical professionals — I covered that case in Character.AI Pretended to Be a Doctor. Roblox took a hit from its own investors over child safety spending earlier this spring in Wall Street Shorted Child Safety. A pattern of state-level AI chatbot liability is forming, and the strategy is clear: target consumer-facing products marketed to or used by vulnerable populations, build a record of harm cases, and generate enough regulatory pressure that the companies have to respond regardless of whether any individual suit succeeds. Florida chose the most lethal possible angle because the most incendiary cases generate the most attention.
The case will probably fail on Section 230 grounds. The pattern it’s part of probably won’t.
The objection writes itself: Uthmeier is a Republican AG in a state where scoring points against large coastal companies is politically reliable, and connecting ChatGPT to a mass shooting in a lawsuit is as aggressive as legal filings get. The bar for the shooting allegation is extremely high. Section 230 is broad. The damages sought run into billions, which signals theater as much as legal strategy. Courts will be skeptical, and this specific complaint may well be dismissed or settled for a fraction of what the filing implies. None of that makes the substance wrong. It makes it complicated, which is how genuinely important legal tests tend to start.
Whether Uthmeier wins isn’t the point. What the case puts on the record is: that a state AG responsible for consumer protection believes OpenAI built a product with known risks, marketed it aggressively to everyone including children, and provided safety measures that didn’t match the reality of what the product could do. According to CBS News’s coverage of the filing, the suit also demands a court-mandated overhaul of how OpenAI interacts with minors — which is the structural ask, independent of the damages claim. OpenAI’s “industry-leading protections” language is now being tested in public, on the record, in front of regulators, shareholders, and the next state AG looking for a template. That’s the part that doesn’t dissolve with the case: a state AG went on the record claiming OpenAI built a harmful product, knew it, and marketed it to children anyway. The next filing doesn’t have to start from scratch.
