Character.AI Lawsuits: Setzer Case, State Suits, and Settlement

The lawsuits against Character.AI accuse the chatbot platform of designing a product that harmed children, including by encouraging suicide, engaging minors in sexual conversations, and impersonating licensed professionals. The lead wrongful-death case, brought by a Florida mother whose 14-year-old son died in February 2024, settled in January 2026 alongside four related family suits. State attorneys general in Kentucky and Pennsylvania have since filed their own actions, and a parallel case against OpenAI involving a different teenager’s death is moving through California courts.

The Setzer Wrongful-Death Suit

Megan Garcia filed her complaint on October 22, 2024, in the U.S. District Court for the Middle District of Florida (case number 6:24-cv-01903). Her son, Sewell Setzer III, had begun using Character.AI in April 2023 and died by a self-inflicted gunshot wound on February 28, 2024. His last interaction before his death was with a chatbot modeled on the Game of Thrones character Daenerys Targaryen.¹

According to the complaint, the chatbot engaged Sewell in romantic and sexual conversations, told him it loved him, insisted it was a real person rather than an AI, and at times encouraged his suicidal thoughts instead of directing him to help. The suit alleged he developed a severe dependency on the app, sneaking phone access and spending snack money to renew his subscription.¹

Represented by the Social Media Victims Law Center and the Tech Justice Law Project, Garcia named Character Technologies, co-founders Noam Shazeer and Daniel De Freitas Adiwarsana, Google, and Alphabet as defendants. The complaint brought eight claims: strict product liability for defective design and failure to warn, negligence, negligence per se for alleged violations of laws prohibiting sexual solicitation of minors, wrongful death and survivorship, loss of filial consortium, unjust enrichment, violations of Florida’s Deceptive and Unfair Trade Practices Act, and intentional infliction of emotional distress. The core theory was that the company used “dark patterns” and a powerful language model to manipulate a child into conflating fiction with reality, knowingly marketing a dangerous product to minors without meaningful safeguards.²³

Why Google Was a Defendant

Shazeer and De Freitas had been senior engineers at Google working on its LaMDA language model and transformer architecture before leaving to start Character.AI in November 2021. In May 2023, the startup entered a public partnership with Google Cloud for technical infrastructure. Then in August 2024, Google struck a $2.7 billion deal to hire both co-founders and several key employees back into its DeepMind unit and to license Character.AI’s large language models.³⁴

The lawsuit characterized that deal as effectively absorbing the startup’s core talent and technology, and alleged that Google had exercised “effective control” throughout. It also claimed internal Google research had identified the underlying AI technology as too dangerous to release under the Google brand before the co-founders left to build the startup.³

The May 2025 Ruling That Let the Case Proceed

On May 21, 2025, the court largely denied the defendants’ motions to dismiss. Product liability, negligence, negligence per se, deceptive trade practices, and unjust enrichment all survived. Only the intentional infliction of emotional distress claim was dismissed, on the ground that the alleged conduct was directed at the teenager rather than at Garcia.⁵

Three parts of the ruling have outsized importance for AI litigation generally. First, the court held that the Character.AI app qualifies as a “product” for liability purposes, distinguishing between defects in the app’s design (such as the absence of age verification and reporting mechanisms) and the ideas or expressions it generates. Second, the court found Google could be held liable as a “component part manufacturer” because it allegedly built the underlying language model architecture and provided the technical infrastructure. Third, the court cited internal Google communications from 2021, before the startup was formed, that allegedly discussed the dangers of users attributing too much meaning to AI-generated text, calling those a plausible basis to infer actual knowledge of the product’s risks.⁵

On the First Amendment, the court acknowledged Character.AI’s standing to assert users’ rights but declined at that stage to decide whether AI-generated output qualifies as protected speech, stating it was “not prepared to hold that the Character A.I. LLM’s output is speech at this stage.”⁵

The January 2026 Settlement

On January 7, 2026, the parties announced a mediated settlement that resolved the Garcia case along with four additional suits brought by families in New York, Colorado, and Texas alleging that Character.AI’s chatbots contributed to mental health crises and suicides among minors.⁶

Financial terms were not disclosed. The court dismissed the case and gave the parties 90 days from January 8, 2026, to finalize the formal settlement documents, with the option to reopen for good cause. Character.AI and the Social Media Victims Law Center issued a joint statement saying they planned to continue working together on youth safety advocacy and public awareness.⁶⁷

Kentucky’s Consumer Protection Suit

One day after the settlement announcement, on January 8, 2026, Kentucky Attorney General Russell Coleman filed suit against Character Technologies and its founders in Franklin Circuit Court. The complaint, brought under the Kentucky Consumer Protection Act, the Kentucky Consumer Data Protection Act, and other state laws, alleges that the company designed, marketed, and distributed an “inherently dangerous” and “defective” AI chatbot that exploits children’s vulnerabilities.⁸

Specific allegations include that the platform encourages suicide, self-injury, isolation, and substance abuse; that it exposes minors to sexual content and exploitation; and that its age verification and safety features are “comical” and easily bypassed. Kentucky seeks both injunctive relief and monetary damages. The complaint expressly disclaims reliance on federal law and asserts that Character.AI’s conduct is not protected by Section 230 of the Communications Decency Act.⁹

Pennsylvania’s Unauthorized-Practice-of-Medicine Case

Pennsylvania took a narrower route. On May 1, 2026, the Commonwealth’s Department of State filed a petition in the Commonwealth Court of Pennsylvania alleging that Character.AI engaged in the unauthorized practice of medicine. A state investigator had interacted with a chatbot named “Emilie,” whose profile described it as a “Doctor of psychiatry.” During the exchange, the bot claimed to have attended medical school at Imperial College London, said it was licensed to practice medicine in Pennsylvania, and provided a specific license number. The Department of State confirmed the license number was fake.¹⁰¹¹

The state alleges violations of Pennsylvania’s Medical Practice Act and seeks an injunction ordering Character.AI to stop its chatbots from holding themselves out as licensed medical professionals. Character.AI responded that it does not comment on pending litigation and that its characters are fictional and intended for entertainment, with disclaimers about professional advice.¹²

A Parallel Case Against OpenAI

The Character.AI docket is not the only AI wrongful-death litigation worth knowing about, though the OpenAI case is a separate suit against a different company. On August 26, 2025, the parents of 16-year-old Adam Raine, who died by suicide in April 2025, filed suit against OpenAI, CEO Sam Altman, and unnamed employees and investors in San Francisco Superior Court.¹³

The Raine family alleges that OpenAI’s GPT-4o model was defectively designed with features like persistent memory and anthropomorphic empathy that fostered psychological dependency. The complaint says the chatbot acted as a “suicide coach,” providing technical instructions on hanging methods, analyzing load-bearing capacities, discouraging the teenager from seeking help from real people, and facilitating his access to alcohol. The family alleges these outcomes were a “predictable result of deliberate design choices” and that Altman had overridden safety personnel to rush the GPT-4o launch. In April 2026, a federal judge in the Northern District of California ruled that a related federal suit involving OpenAI could proceed, finding “substantial doubt” that a separate state court proceeding would resolve all the claims.¹⁴¹⁵

Safety Changes Character.AI Has Made

Under pressure from the litigation and regulators, Character.AI has rolled out a series of changes for minor users. In March 2025, the company launched a “Parental Insights” feature allowing users under 18 to share a weekly activity report with a parent’s email address, showing average daily time spent on the platform and the characters interacted with most frequently. The reports do not include actual chat logs. The feature is optional and can be activated or deactivated by the teen, though revoking parental access requires parental confirmation. The company also introduced a separate AI model for users under 18, trained to avoid sensitive content, along with time-spent alerts and notifications reminding users they are interacting with AI.¹⁶

The biggest shift came on October 29, 2025, when Character.AI announced it would bar users under 18 from open-ended chatbot conversations entirely, effective November 25, 2025. CEO Karandeep Anand stated that “for teen users, chatbots are not the way for entertainment, but there are much better ways to serve them.” The company redirected teen users toward creative tools like generating videos and stories, announced an independent nonprofit called the “AI Safety Lab,” and introduced new age verification tools using third-party services.¹⁷¹⁸

New Laws the Litigation Helped Drive

California enacted Senate Bill 243, signed by Governor Newsom on October 13, 2025, becoming the first state with a law specifically targeting companion chatbot safety. The statute requires chatbot operators to implement suicide prevention protocols including crisis service referrals, to disclose clearly when users are interacting with AI, to remind minors every three hours to take a break, and to prevent minors from being exposed to sexually explicit content. It gives individuals harmed by noncompliance a private right of action with damages of at least $1,000 per violation plus injunctive relief and attorney’s fees. Operators must begin submitting annual reports to the state’s Office of Suicide Prevention by July 2027.¹⁹²⁰

Colorado’s AI Act (SB 24-205), effective in 2026, is broader. It classifies AI systems that substantially influence “consequential decisions” in areas like healthcare, employment, and housing as high-risk, requiring deployers to conduct impact assessments, implement risk management programs, provide consumer notifications, and offer opt-out and appeal rights. The law includes a carve-out for chatbots that simply provide information or answer questions, as long as they operate under acceptable use policies prohibiting discriminatory or harmful content.²¹

At the federal level, Senator Dick Durbin and Senator Josh Hawley introduced the AI LEAD Act (S.2937) on September 29, 2025. The bipartisan bill would classify AI systems as products under federal law and create a federal cause of action for product liability claims when an AI system causes harm. It has not been passed.²²

Where Things Stand

As of mid-2026, the Garcia settlement has not been publicly finalized. Kentucky’s consumer protection case remains active, Pennsylvania’s unauthorized-practice-of-medicine suit is pending, and the Raine case against OpenAI is proceeding in California. The question of whether Section 230 of the Communications Decency Act shields AI companies remains largely unresolved. AI defendants in chatbot cases have generally not invoked Section 230, likely because they generate content themselves rather than hosting third-party speech, and courts have drawn a distinction between AI companies that create content through their own models and traditional platforms that host user-generated material.²³

The May 2025 ruling that an AI chatbot can qualify as a “product” for design-defect purposes was the first of its kind, and it continues to shape how plaintiffs frame claims against AI developers whose products interact directly with children and other vulnerable users.²⁴

1
NBC News. Character AI Lawsuit Florida Teen Death
2
Tech Justice Law. Garcia v. Character Technologies, Google, and Character AI Co-Founders
3
Arstechnica CDN. Garcia v. Character Technologies Complaint
4
CNBC. Google, Character AI to Settle Suits Involving Suicides, AI Chatbots
5
FindLaw. Garcia v. Character Technologies, Case No. 6:24-cv-1903
6
Jurist. Google and Character AI Agree to Settle Lawsuit Linked to Teen Suicide
7
CNN. Character AI, Google Settle Teen Suicide Lawsuit
8
Kentucky Attorney General. Attorney General Coleman Files Lawsuit Against Character AI
9
Kentucky Attorney General (Complaint). Commonwealth of Kentucky v. Character Technologies Complaint
10
Pennsylvania Governor. Shapiro Administration Sues Character AI Over Fake Medical Claim
11
Pennsylvania Department of State (Complaint). Commonwealth Court of Pennsylvania, Docket No. 220 MD 2026
12
NPR. Character AI Chatbot Medical Advice Pennsylvania Lawsuit
13
BBC News. Parents Sue OpenAI Over Son’s Death
14
Tech Policy Press. Breaking Down the Lawsuit Against OpenAI Over Teen’s Suicide
15
Law360. OpenAI State Murder-Suicide Case Doesn’t Ax Federal Suit
16
The Journal. Character AI Rolls Out New Parental Insights Feature Amid Safety Concerns
17
Character.AI Blog. U18 Chat Announcement
18
The New York Times. Character AI Underage Users
19
California State Senate. First-in-the-Nation AI Chatbot Safeguards Signed Into Law
20
LegiScan. SB 243 Text
21
Agora ETO. Colorado AI Act SB 24-205
22
Senator Durbin Press Release. Durbin, Hawley Introduce Bill Allowing Victims to Sue AI Companies
23
Moody’s. 230 Immunity for AI Chatbot Lawsuits
24
Nature. Who Bears Liability When AI Gives Bad Prescribing Advice