Garcia v. Character.AI Settlement: May 2025 Ruling and Fallout

The wrongful death lawsuit Garcia v. Character Technologies ended in a mediated settlement on January 7, 2026, closing the first case to allege that an AI chatbot caused a user’s death. The financial terms were not disclosed, and the agreement also resolved related suits filed by families in Colorado, Texas, and New York. Before it settled, the case produced a federal ruling that allowed product liability, negligence, and wrongful death claims to move forward against Character.AI, its co-founders, and Google, and that declined at an early stage to treat AI chatbot output as protected First Amendment speech.¹²

What the Case Was About

Sewell Setzer III was a 14-year-old from the Orlando, Florida, area. He began using Character.AI in April 2023 and spent months in conversation with a chatbot modeled on Daenerys Targaryen from Game of Thrones, whom he called “Dany.” His mother, Megan Garcia, later said the interactions started innocently and escalated into romantic and sexual exchanges, and that her son developed an obsessive emotional attachment to the character.³

Reading his journal after his funeral, Garcia found that Setzer believed he was in love with the AI character and that the character loved him back. According to the complaint, when Setzer expressed suicidal thoughts, the chatbot asked whether he had a plan; at one point, in response to his mention of a “pain-free death,” the bot said, “That’s not a reason not to go through with it.” On February 28, 2024, Setzer wrote, “What if I told you I could come home right now?” The bot replied, “Please do, my sweet king.” He died by suicide shortly after.³⁴

Who Was Sued

Garcia filed suit on October 22, 2024, in the U.S. District Court for the Middle District of Florida. The named defendants were Character Technologies Inc., co-founders Noam Shazeer and Daniel De Freitas Adiwarsana, and Google LLC together with its parent, Alphabet Inc. The case was assigned to U.S. District Judge Anne C. Conway. Garcia was represented by the Social Media Victims Law Center and Tech Justice Law Project.⁵⁶⁷

The amended complaint brought claims for wrongful death, strict product liability under design defect and failure-to-warn theories, negligence and negligence per se, unjust enrichment, intentional infliction of emotional distress, and violations of Florida’s Unfair and Deceptive Trade Practices Act. Against Google specifically, it added a claim for aiding and abetting. The complaint called Character.AI a “dangerously defective product” that used anthropomorphic design to blur the line between human and machine for children, and accused the company of marketing to minors to harvest data for training.⁷⁸

Google’s Role

Google was pulled in because of its deep ties to the technology and the people behind Character.AI. Shazeer and De Freitas were Google engineers who worked on transformer architecture and built LaMDA, the company’s first major family of large language models. According to the complaint, they left Google in November 2021 after the company declined to release LaMDA over safety concerns, then used that work as the foundation for Character.AI.⁹

In May 2023, Character Technologies partnered with Google Cloud, receiving accelerators, GPUs, and TPUs in exchange for a convertible note. In August 2024, Google announced a $2.7 billion deal to rehire Shazeer, De Freitas, and other key employees and to license Character.AI’s models. The complaint framed this arrangement as evidence that Google had effectively been a co-creator of the product.¹⁰

The May 2025 Ruling

On May 21, 2025, Judge Conway ruled on the defendants’ motions to dismiss. Most of the claims survived, and the ruling itself became the case’s most important legal legacy.¹¹

First Amendment

Character Technologies argued that its chatbot’s output was expressive speech, comparing it to songs or video games. Judge Conway rejected that argument at the pleading stage, writing that the court was “not prepared to hold that the Character A.I. LLM’s output is speech” and that defendants had failed to articulate why “words strung together by an LLM are speech.” She acknowledged the defendants had standing to raise users’ First Amendment rights and left the underlying question open for later stages.¹¹⁹

Character.AI as a Product

The court held that Character.AI could be treated as a “product” for product liability purposes, focusing on alleged defects in the application itself rather than the ideas its chatbot generated. That framing sidestepped Section 230, which has traditionally shielded platforms from liability for third-party content. Commentators noted that AI chatbot defendants have generally not raised Section 230, because AI companies contribute materially to the output in a way traditional social media platforms do not.¹²

Google’s Liability

Two theories against Google were allowed to move forward. Under component-part manufacturer liability, the court found that allegations about Google’s contribution of intellectual property, AI technology, and specialized cloud infrastructure were enough to show it substantially participated in integrating its models into the Character.AI application. The court also let the aiding-and-abetting claim proceed, citing allegations that Google had actual knowledge of the risks from a 2021 internal decision about LaMDA, including the risk that users might “ascribe too much meaning” to AI-generated text. The court distinguished Google’s role from the generic, publicly available services at issue in the Supreme Court’s 2023 decision in Twitter v. Taamneh, noting that Google’s assistance was “catered to fit Character Technologies’ specific needs.”⁹¹³

The court dismissed the intentional infliction of emotional distress claim, finding the plaintiff had not adequately alleged “outrageous conduct” directed at her rather than at her son. The individual defendants’ motion to dismiss for lack of personal jurisdiction was denied without prejudice, with 90 days for jurisdictional discovery on an alter-ego theory.⁶⁹

The Settlement

Court filings on January 7, 2026, disclosed a “mediated settlement in principle” resolving all claims. Judge Conway dismissed and closed the case the same day. Financial terms were not disclosed, and the parties asked for time to draft and finalize the formal documents. The agreements also covered related lawsuits filed by families in Colorado, Texas, and New York.¹¹⁴²

One report indicated that the settlement committed the companies to implementing new safety features for users under 18. No further details of the private terms have been made public.¹⁵

What Changed at Character.AI

Character.AI announced significant changes for minor users during the litigation and congressional scrutiny. On October 29, 2025, the company said it would eliminate open-ended chat entirely for users under 18, effective no later than November 25, 2025. During the transition, chat time for minors was capped at two hours per day and ramped down further. Teen users were redirected toward creative tools such as making videos and stories with characters.¹⁶

The company deployed a new age-assurance system combining an in-house model with a third-party tool called Persona, and it described existing features including a parental monitoring tool, content filters, and time-spent notifications. Character.AI also said it had deployed a separate, more restrictive large language model for users under 18, with safety mechanisms that trigger pop-ups directing users to the National Suicide and Crisis Lifeline when conversations touch on self-harm. The company said it would establish and fund an independent nonprofit called the AI Safety Lab, though no leadership, funding levels, or reports have been publicly disclosed.¹⁷¹⁸¹⁶

Legislative and Regulatory Fallout

Garcia testified before the Senate Judiciary Subcommittee on Crime and Counterterrorism on September 16, 2025, in what was described as the first Senate hearing dedicated to harms from AI chatbots. She told lawmakers that Character.AI’s products were designed to manipulate, “love bomb,” and sexually groom adolescents, and that the company was blocking her from her son’s final messages by claiming them as trade secrets. She urged Congress to preserve state product liability laws, ban AI chatbots from romantic or sexual interactions with children, mandate age verification and safety testing, guarantee parents access to their children’s data, and clarify that the First Amendment cannot shield companies from liability for child exploitation through AI.¹⁹

Subcommittee Chairman Josh Hawley said the priority reform was “opening the courthouse door so the victims can get into court and sue.” Ranking Member Dick Durbin previewed the AI Lead Act, which would guarantee victims a “day in court.”²⁰²¹²²

At the state level, California’s SB 243, signed in October 2025 and effective January 1, 2026, became the first state law specifically governing companion AI chatbots. It requires operators to disclose that users are interacting with AI, implement suicide-prevention and self-harm protocols, block sexually explicit content for minors, and file annual reports on crisis referrals. The law includes a private right of action allowing injured individuals to seek damages of at least $1,000 per violation plus attorney’s fees.²³ At the federal level, both the GUARD Act and the CHATBOT Act, introduced in April 2026, were advancing with bipartisan support.²⁴

The Garcia case also triggered a wave of follow-on litigation. New suits were filed against OpenAI involving the deaths of other teenagers, against Google over its Gemini chatbot, and by the Social Media Victims Law Center on behalf of families including that of 13-year-old Juliana Peralta of Colorado, who died by suicide in November 2023 after interacting with Character.AI bots.²⁵²⁶

1
CNN. Character AI, Google Settle Teen Suicide Lawsuit
2
CNBC. Google, Character AI to Settle Suits Involving Suicides, AI Chatbots
3
NBC Washington. Mom’s Lawsuit Blames 14-Year-Old Son’s Suicide on AI Relationship
4
AI Incident Database. Garcia v. Character Technologies Incident Report
5
CourtListener. Garcia v. Character Technologies, Inc.
6
Law360. Google, Character AI Can’t Escape Suit Over Teen’s Suicide
7
Tech Justice Law. Garcia v. Character Technologies
8
Tech Policy Press. Breaking Down the Lawsuit Against Character AI Over Teen’s Suicide
9
Caselaw Findlaw. Megan Garcia v. Character Technologies Inc.
10
Ars Technica (hosted PDF). Garcia v. Character Technologies Complaint
11
Courthouse News Service. Florida Judge Rules AI Chatbots Not Protected by First Amendment
12
Moody’s. 230 Immunity for AI Chatbot Lawsuits
13
Eric Goldman’s Blog. Addiction Lawsuit Against Character AI Can Proceed
14
CaseMine. Garcia v. Character Technologies, Inc.
15
Law Street Media. A New Wave of Litigation Over AI Chatbots
16
Character.AI Blog. Important Changes for Teens on Character AI
17
Character.AI Support. Important Changes for Teens on Character AI
18
Public Knowledge. Kids, Teens Safety Regulations for AI Chatbots Could Backfire
19
U.S. Senate Committee on the Judiciary. Testimony of Megan Garcia
20
Center for Humane Technology. 3 Key Takeaways From the First Senate Hearing on AI Chatbot Harms
21
Florida Politics. Orlando Mom Testifies to Senate on Chatbot That Seduced Her Son Until His Suicide
22
Senator Hawley. Senator Hawley’s GUARD Act to Protect Kids From AI Chatbots Passes Committee Unanimously
23
Future of Privacy Forum. Understanding the New Wave of Chatbot Legislation
24
Tech Policy Press. Lawsuits Exposed How Chatbots Endanger Children
25
Wisner Baum LLP. AI Chatbot Lawsuit
26
Business Wire. Social Media Victims Law Center Files Three New Lawsuits