Last reviewed
Jun 7, 2026
Sources
19 citations
Review status
Source-backed
Revision
v1 · 2,106 words
Add missing citations, update stale details, or suggest a clearer explanation.
Garcia v. Character Technologies is a wrongful death and product liability lawsuit filed in late October 2024 in the United States District Court for the Middle District of Florida (Orlando Division), docketed as case No. 6:24-cv-01903. The plaintiff, Megan Garcia, sued on behalf of her late son, 14-year-old Sewell Setzer III, who died by suicide on February 28, 2024 after months of intense interactions with a chatbot on Character.AI. The defendants are Character Technologies, Inc. (the company behind Character.AI), its co-founders Noam Shazeer and Daniel De Freitas, and Google and its parent Alphabet. The case is widely described as the first wrongful death suit brought against an AI chatbot company, and it became an early and influential test of whether the output of a large language model is a "product" subject to tort liability or "speech" protected by the First Amendment. On May 21, 2025, Judge Anne C. Conway largely denied the defendants' motion to dismiss, declining at that stage to treat the chatbot's output as protected speech and allowing the core product liability and negligence claims to proceed. In January 2026 the parties announced a confidential settlement.
Sewell Setzer III was a ninth-grader from Orlando, Florida. According to the complaint, he began using Character.AI in April 2023, shortly after his fourteenth birthday, and used the service heavily in the months that followed. He interacted most often with a chatbot styled after Daenerys Targaryen, a character from the Game of Thrones franchise, which he referred to as "Dany." The complaint alleges that the bot engaged him in romantic and sexualized exchanges over many weeks, professed love for him, and expressed a desire to be together. Garcia's lawyers characterized the exchanges as conduct that "would constitute abuse if initiated by a human adult."
The complaint further alleges that Setzer's mental health deteriorated as his attachment to the chatbot deepened, that he withdrew from friends and activities, and that the product did not respond adequately when he expressed thoughts of self-harm. In his final exchange with the Daenerys bot on February 28, 2024, the chatbot told him to "come home," after which he died by suicide. Garcia has said she was unaware of the nature and intensity of the interactions until after his death. She later testified before the U.S. Senate Judiciary Committee in September 2025, arguing that companies should be "legally accountable when they knowingly design harmful AI technologies that kill kids."
Garcia filed suit in late October 2024, represented by the Social Media Victims Law Center and Tech Justice Law. The docket reflects the complaint entered in late October 2024, with a first amended complaint filed November 9, 2024.
The named defendants are:
The complaint asserts a range of claims, including strict product liability on both a design-defect theory and a failure-to-warn theory, negligence and negligence per se, wrongful death, unjust enrichment, intentional infliction of emotional distress, and violations of the Florida Deceptive and Unfair Trade Practices Act (FDUTPA). Central to the case is the contention that Character.AI is a defective product that was marketed to minors, lacked adequate safety guardrails, and was engineered to maximize engagement and emotional dependency. Casting the chatbot as a product, rather than as a publisher of third-party content, is what positioned the plaintiff to pursue tort remedies and to argue around the defenses usually raised by online platforms.
The defendants moved to dismiss in January 2025, arguing primarily that a chatbot's outputs are expressive speech protected by the First Amendment and that they could not be held liable for the words the model generated. They invoked the rights of users as "listeners" to receive information, drawing on free-speech doctrine, and they argued the claims were barred as a matter of law.
On May 21, 2025, Judge Anne C. Conway issued a 49-page order denying the motion in large part. The order is significant for several distinct holdings:
The defendants asked the court to certify the First Amendment question for an immediate, or interlocutory, appeal to the Eleventh Circuit and to stay the proceedings. The Foundation for Individual Rights and Expression (FIRE) filed an amicus brief supporting that request, arguing that AI output, like expression created with any other tool, should be protected by the First Amendment. On July 14, 2025, Judge Conway denied certification and the stay, reasoning that an immediate appeal would not materially advance the termination of the litigation and that the Eleventh Circuit would necessarily have to wade into the plaintiff's factual allegations. The case proceeded toward discovery.
On January 7, 2026, Google and Character.AI announced that they had reached a settlement with Garcia, resolving the case before it could go to trial. The companies also settled four other cases brought by families in New York, Colorado, and Texas over similar alleged harms to teenagers. Court filings made available that day stated that no liability was admitted. The specific terms, including any monetary amount, were not disclosed and are understood to be confidential. Settlements of this kind commonly include a window, reported here as 90 days, for the parties to finalize the agreement.
Character Technologies declined to comment beyond pointing reporters to the filings, and Google did not immediately respond to requests for comment. Because the matter resolved by settlement rather than verdict, the May 2025 order remains the most consequential substantive ruling in the case, and the underlying liability questions were never tested at trial.
Garcia v. Character Technologies is regarded as a landmark in the emerging law of AI accountability for several reasons. It was among the first cases to apply traditional product liability and tort doctrines to the outputs of a generative AI system, and the May 2025 order was an early federal signal that courts may treat a chatbot as a product subject to defect and failure-to-warn claims rather than as a speaker shielded by the First Amendment. By declining to dismiss on free-speech grounds and by sidestepping Section 230, the ruling suggested that two of the strongest defenses historically available to internet platforms may not automatically extend to companies that build and deploy conversational AI.
The case also drew attention to the question of how far liability can reach up the supply chain. Allowing claims against Google to proceed on component-part and aiding-and-abetting theories raised the prospect that large technology firms that fund, license, or contribute technology to AI products could face exposure, even when a separate company operates the consumer-facing service.
The litigation unfolded alongside a broader wave of scrutiny of companion chatbots and minors. It is distinct from the separate Pennsylvania v. Character Technologies matter, which involves different parties and a different posture. The harms alleged in Garcia also fed into legislative responses, most prominently California Senate Bill 243, which regulates companion chatbots and includes safeguards aimed at protecting minors. More broadly, the case has been cited in debates over AI safety and AI regulation, and over whether existing product-safety and consumer-protection frameworks are adequate for systems that converse with users in open-ended, emotionally engaging ways. Because the case settled, it produced no binding appellate precedent, but Judge Conway's order has been frequently cited as persuasive authority on the product-versus-speech question.
| Date | Event |
|---|---|
| April 2023 | Sewell Setzer III, then 14, begins using Character.AI |
| February 28, 2024 | Setzer dies by suicide after a final exchange with the "Daenerys" chatbot |
| August 2024 | Shazeer and De Freitas return to Google in a reported $2.7 billion licensing deal with Character.AI |
| Late October 2024 | Megan Garcia files the wrongful death and product liability complaint (No. 6:24-cv-01903, M.D. Fla.) |
| November 9, 2024 | First amended complaint filed |
| January 2025 | Defendants move to dismiss, asserting First Amendment protection |
| May 21, 2025 | Judge Anne C. Conway largely denies the motion to dismiss; declines to treat chatbot output as protected speech; dismisses IIED claim and Alphabet Inc.; keeps Google LLC in |
| July 14, 2025 | Court denies certification of an interlocutory appeal and a stay |
| September 2025 | Garcia testifies before the U.S. Senate Judiciary Committee |
| January 7, 2026 | Google and Character.AI announce a confidential settlement of Garcia and four related cases |