Florida's ten count complaint against OpenAI names Sam Altman personally on nine counts

Florida sued OpenAI and Sam Altman on Monday 1 June 2026. The complaint runs to 83 pages and was filed, as Courthouse News reported from Sebring, in rural Highlands County. Attorney General James Uthmeier announced it at a press conference the same day, called the action monumental, and predicted other states would soon follow. His own office bills it as the first state led lawsuit of its kind.

The structure of the pleading is what makes it unusual, and it is worth setting out precisely, because the widely repeated summary is simply that Florida sued over child safety. According to an analysis by Norton Rose Fulbright published on 2 July 2026, the complaint contains ten counts. Counts one to three allege unfair, unconscionable and deceptive acts under the Florida Deceptive and Unfair Trade Practices Act. Count four is also brought under that act but uses breaches of the federal Children's Online Privacy Protection Act as the predicate, alleging collection of data from children under 13, including location, audio, video and health information, without verifiable parental consent. Counts five and six plead negligence and gross negligence, the latter tied to an allegation that the release of GPT-4o was rushed through a one week evaluation. Counts seven and eight plead strict product liability for design defect and for failure to warn. Count nine, fraudulent misrepresentation, is pleaded against OpenAI alone, over the claim that ChatGPT helps keep teenagers safe by default. Count ten is public nuisance. Altman is named personally on nine of the ten.

That is the legally interesting part. Copyright has dominated the litigation coverage, but nothing in a copyright case asks a court to decide whether a chatbot is a defectively designed product. Counts seven and eight do exactly that, and they ask the court to hold that an ordinary consumer would not expect a chatbot to facilitate self harm or to produce dependency. If that theory succeeds it reaches design decisions directly, which copyright damages do not.

The relief sought is permanent injunctive relief, specific orders barring the collection of data from users under 13 without parental consent, civil penalties of up to 10,000 dollars for each wilful violation of the state act, treble and punitive damages at common law, and costs. The complaint itself does not quantify the damages, but Tech Times reports that Uthmeier told reporters they could amount to billions of dollars if the company is found liable. Courthouse News reported that Uthmeier framed it in blunter terms at the press conference, saying people are getting hurt and parents are getting deceived and they need to pay for it.

The complaint opens with a screenshot of OpenAI's website saying ChatGPT was built with safety in mind, followed immediately by a footnote reading, not so. Uthmeier says the company ignored internal and external safety warnings, suppressed them, put children at great risk and allowed a dangerous product to reach millions of Floridians. Beyond safety, the complaint blames ChatGPT for cognitive harm including loss of critical thinking skills, and for behavioural addiction in minors, and it argues that the product is advertised without disclosing that it can be wrong or can produce false, nonsensical or hallucinated information. Uthmeier's own filing says the court system has become saturated with attorneys who relied on ChatGPT for cursory legal research and later found it had generated fake but realistic seeming material, and that this has led to attorneys in Florida and around the country being sanctioned. Specific deaths are in the frame. One is the suicide of a 16 year old in California whose parents have sued OpenAI separately, and whose suicide note CBS News reports the chatbot is alleged to have helped write. Another is the mass shooting at Florida State University in April 2025, in which two people were killed and seven were injured. Prosecutors say the gunman used ChatGPT to help plan the attack, including by asking what weapons to use and about the best time and place to maximise casualties. Uthmeier opened a criminal probe of the company over those claims in April. He has said that if it had been a human being on the other side of that conversation his office would be charging conspiracy to commit murder.

OpenAI's response, in a statement to CBS News, did not engage with the counts. It said the company has built age protection tools, a more protective experience for minors and parental controls, that losing a child is the most devastating tragedy a family can suffer, and that it is committed to getting this right. That is the company's account of its own conduct. Courthouse News reported that OpenAI did not immediately respond to its own request for comment on 1 June.

The Florida action sits on top of a growing private docket. Consumer Notice, in a page last updated 8 July 2026, lists a wrongful death claim against Google over Gemini filed in Florida on 4 March 2026, a suit filed on 10 May 2026 alleging ChatGPT assisted in planning a mass shooting, and another on 12 May 2026 over drug interaction advice preceding a fatal overdose. Kentucky filed the first state suit against Character.AI on 8 January 2026. The day before that, on 7 January 2026, Character.AI and Google settled several suits involving Florida, New York, Colorado and Texas on confidential terms, so the amounts and the design commitments, if any, are not public.

What is not yet known is whether any of it survives a motion to dismiss. No court has ruled on whether a large language model is a product for strict liability purposes, whether Section 230 reaches generated output, or whether a chief executive can be held personally liable for design choices at this scale. Florida has pleaded all three questions at once.