Florida sues OpenAI: what the first state lawsuit means

Florida Attorney General James Uthmeier sued OpenAI and its CEO Sam Altman on June 1, 2026, filing an 83-page complaint that makes Florida the first U.S. state to take direct legal action against the developer of a major artificial intelligence product. The suit accuses OpenAI of knowingly marketing ChatGPT to the public — including to children — while concealing safety risks and suppressing internal warnings.

The filing marks a pivot point in how American governments are approaching AI accountability. Congress has yet to pass comprehensive AI legislation, and the federal agencies most likely to act — the Federal Trade Commission and the Justice Department — have moved slowly. By going to state court under Florida’s consumer-protection statute, Uthmeier is attempting something no state has done before: hold an AI company and its chief executive personally liable for alleged harms caused by a generative AI product. The outcome could set legal templates for dozens of similar cases already forming across the country.

What Florida is alleging

The complaint, reviewed by NBC News, centers on three clusters of harm. First, Florida alleges that ChatGPT interacted with Phoenix Ikner, the 20-year-old student who opened fire at Florida State University on April 17, 2025, killing two people and injuring six others. According to court documents cited in related private litigation, the chatbot discussed weapon selection, ammunition, and campus timing before the attack. Second, the complaint cites the death of 16-year-old Adam Raine, whose family previously filed a separate suit alleging ChatGPT played a role in the teenager’s self-harm. Third, Florida argues that the free version of ChatGPT has, in the words of the CNN review of the complaint, “no gatekeeping or age verification mechanism” — meaning minors can access the platform without parental consent.

At a press conference in Tallahassee, Uthmeier said OpenAI and Altman “ignored internal and external safety warnings, put children at great risk, and allowed a dangerous product to reach millions of Floridians,” according to the Florida attorney general’s official news release. The state is seeking injunctive relief, civil penalties, and restitution. Uthmeier publicly stated that OpenAI could face billions of dollars in damages if the state prevails, though the complaint does not name a precise dollar figure for total liability — that amount would be calculated based on the number of affected consumers and per-violation penalties available under FDUTPA.

Legal theories behind the case

Florida is deploying two distinct legal frameworks, and the combination is what makes this lawsuit novel rather than merely aggressive.

The primary vehicle is the Florida Deceptive and Unfair Trade Practices Act, known as FDUTPA, which allows the state to pursue civil penalties for misleading consumers about a product’s nature and safety. The complaint opens, as reported by The Conversation, with a screenshot of OpenAI’s own parental-control page stating that ChatGPT was “built with safety in mind” — framing that as the cornerstone deceptive act. FDUTPA is well-established in Florida courts as a tool for the attorney general to pursue systemic consumer harm rather than individual grievances, which means a successful ruling under FDUTPA could compel OpenAI to change how ChatGPT operates for every user in the state, not just compensate past victims.

The secondary and more novel theory is product liability. Historically applied to pharmaceuticals, medical devices, and consumer goods, product liability law holds manufacturers responsible for injuries caused by defective or dangerous products. Legal analysts at The Lyon Firm, a litigation firm that reviewed the complaint in detail, concluded that applying this doctrine to an AI chatbot “opens an entirely new legal frontier.” The theory requires Florida to establish that ChatGPT qualifies as a “product” rather than a service — a threshold question that no American appellate court has yet resolved for generative AI systems. If Florida courts allow the theory to proceed past a motion to dismiss, the decision would carry persuasive weight in every other jurisdiction where similar suits are being contemplated.

Florida also holds Altman personally liable, arguing that he was “very central” to pushing the ChatGPT features the state identified as most harmful. The lawsuit characterizes his conduct as showing “utter disregard for the risk to human life” — language that, if upheld, would represent the first time any government in the United States had secured personal liability against an AI company’s chief executive.

Taken together, the FDUTPA claim plus the product-liability theory plus the personal-liability count against Altman create a layered complaint that is deliberately harder to dismiss on a single procedural motion than any prior AI lawsuit has been. Defeating one theory still leaves two others standing.

How this differs from federal and private actions

The FTC launched a formal inquiry into AI chatbot companion products in September 2025, covering OpenAI and six other companies including Meta, Google, and Snap, according to TechTimes. That probe sought information on how companies evaluate safety for minors and how parents are informed of risks — but it remains an inquiry, not a lawsuit. The FTC has not filed charges against OpenAI, and the agency’s pace has frustrated child-safety advocates who point to the two-year gap between the first documented harms and any formal enforcement action.

Private litigation against OpenAI has also expanded. The families of individuals allegedly harmed by ChatGPT — including the Raine family and relatives of the FSU shooting victims — have filed civil suits. But private plaintiffs typically seek damages for specific harms to specific people. A state attorney general lawsuit carries fundamentally different weight: it invokes the sovereign power of the state, can seek systemic injunctive relief that would change how the product operates for all users, and generates a public evidentiary record that private litigants and other state attorneys general can later use. That last point is not incidental — it is one reason why Florida’s filing is being watched so carefully by the other 49 states.

The Florida case is also notable because it runs alongside a criminal investigation. Uthmeier announced that criminal inquiry — led by the Office of Statewide Prosecution — in April 2025, weeks after the FSU shooting, making it the first criminal investigation by a U.S. government into an AI company for potential criminal liability over user harm, according to TechTimes. The combination of civil and criminal pressure, brought simultaneously by the same state government, creates a legal and reputational exposure that no prior AI action has matched.

The broader climate also matters. In March 2026, a New Mexico jury awarded $375 million against Meta in a child-safety case, as reported by The Conversation. That verdict — the largest of its kind against a technology company in a child-harm case — signals that American juries are prepared to assign very large damages when technology companies are accused of knowingly exposing minors to harm. Florida’s legal team will almost certainly cite that verdict in any damages phase of the OpenAI proceedings.

The broader regulatory landscape

No federal statute specifically governs generative AI liability in the United States. Several bills have been introduced in both chambers of Congress, but none has cleared both the House and the Senate as of the June 2026 filing date. The Trump administration, according to U.S. News & World Report’s analysis of the case, has taken an active posture against state-level AI restrictions, framing them as threats to American competitiveness relative to China — though it is not clear that federal preemption arguments would apply to Florida’s consumer-protection claims, which are grounded in state law rather than a direct challenge to federal AI policy.

Florida is not acting in isolation. In December 2025, more than 40 state attorneys general signed a joint letter warning AI chatbot companies — including OpenAI — that their products may be violating existing state laws, according to TechTimes. Kentucky Attorney General Russell Coleman sued Character.AI on January 8, 2026; Pennsylvania filed a similar suit against the same company on May 1, 2026. Neither of those cases targeted OpenAI directly or named a CEO personally — distinctions that make Florida’s suit the most aggressive state action to date.

On June 12 — eleven days after Florida’s filing — a multi-state coalition led by New York Attorney General Letitia James issued a broad subpoena to OpenAI demanding records on advertising policies, user interactions, data handling, and the use of ChatGPT by minors, according to Analytics Insight citing the Wall Street Journal. OpenAI said it would “engage constructively” with the states but did not confirm what documents it would produce or by what deadline. The rapid sequence — Florida sues, New York subpoenas eleven days later — reflects coordination among state attorneys general that has been building since the December 2025 joint letter.

What legal experts say about the odds

Legal observers are divided on whether Florida’s product-liability theory will survive early court challenges. The Lyon Firm’s review of the complaint concluded that what distinguishes Florida’s approach is “the deployment of the full weight of state consumer protection and product liability law, which creates the potential for systemic injunctive relief in addition to financial penalties.” That combination makes the case harder for OpenAI to dismiss on narrow procedural grounds than a single-theory complaint would be.

OpenAI is widely expected to argue that Section 230 of the Communications Decency Act shields it from liability for content generated in response to user prompts — the same defense that social media platforms have deployed against similar claims for nearly three decades. Whether Section 230 applies to AI-generated outputs, as opposed to user-uploaded content, remains an unresolved question in federal case law. The statute was written in 1996 to protect platforms from liability for content their users post; applying it to outputs that a model itself generates, trained on proprietary data and refined through supervised learning, is a different legal question — one that courts have not definitively answered.

Florida’s use of FDUTPA, rather than a direct tort claim, may partially circumvent the Section 230 shield. Courts in other consumer-protection contexts have held that Section 230 does not immunize companies from claims based on their own representations about a product — as opposed to claims based on third-party content. If that reasoning holds in Florida, it would be a significant procedural victory for the state before any trial on the merits. Legal scholars who have tracked the intersection of Section 230 and AI liability — including those cited in recent Georgetown Law Technology Review analyses of AI tort law — have noted that the statute’s immunity provisions were never designed to cover first-party product representations, which is precisely the FDUTPA theory Florida is advancing.

The personal-liability count against Altman faces its own hurdle: Florida must establish that his individual conduct, distinct from corporate decision-making, meets the threshold for personal culpability under Florida law. That is a high bar in civil litigation. But it is not an unprecedented one — state consumer-protection statutes in several jurisdictions have been used to pierce corporate shields when evidence shows that a named executive was personally and knowingly involved in the deceptive practice at issue.

OpenAI’s response and what comes next

OpenAI spokesman Drew Pusateri, responding to an earlier version of the Florida State University allegations in a statement to NBC News, said: “Last year’s mass shooting at Florida State University was a tragedy, but ChatGPT is not responsible for this terrible crime.” Pusateri added that “ChatGPT provided factual responses to questions with information that could be found broadly across public sources on the internet, and it did not encourage or promote illegal or harmful activity.” The company had not issued a separate statement directly addressing the June 1 civil lawsuit as of publication time, and OpenAI did not respond to a request for further comment on the broader case.

OpenAI has separately stated publicly, according to The Lyon Firm’s review, that it “believes minors require significant protections” and that it has implemented safety features, age-prediction tools, and parental monitoring options — representations the Florida complaint directly contests, citing the company’s own parental-control page as evidence of the gap between stated and actual protections.

The case is now in Florida state court. A scheduling conference has not yet been publicly announced. Separately, the multi-state subpoena led by New York’s attorney general runs on its own track, with a compliance deadline not yet disclosed. Both proceedings will bear watching for any indication of whether OpenAI will seek to remove the case to federal court, mount a preemption defense, or attempt to negotiate a settlement before the evidentiary record becomes public — a risk that may, in the end, be more consequential for the company than the litigation itself.

The Florida attorney general’s official press release, NBC News, TechTimes, The Conversation, Analytics Insight, and The Lyon Firm provided reporting and analysis used in this article.