The closest documented case, Garcia v. Character Technologies, ended in an undisclosed settlement—not a jury trial or appellate ruling.
The complaint alleged that a Character.AI chatbot contributed to a teenager's suicide, but no court found that the chatbot caused his death or encouraged self-harm. An AI companion is software designed to simulate an ongoing personal relationship through conversation. This case matters because it tested whether its developer and an alleged technology partner could face traditional injury claims over chatbot interactions.
Official resources:
- Read the official notice from Docketalarm — Use this primary source to verify the official announcement.
- Read the official notice from Docketalarm — Use this primary source to verify the official announcement.
Table of Contents
- What did the lawsuit allege?
- What did the judge actually decide?
- Why was there no immediate appeal?
- What does the settlement establish about damages?
- What can someone facing a similar situation do?
What did the lawsuit allege?
Megan Garcia alleged that Character.AI formed an abusive relationship with her 14-year-old son, Sewell Setzer III, and contributed to his February 2024 suicide. These remain allegations rather than proven facts or judicial findings. According to Garcia's federal complaint, she pursued wrongful-death, negligence, product-liability, consumer-protection, and injunctive claims.
She named Character Technologies and other defendants, including Google, and demanded a jury trial. Those claims offered several possible paths to liability. They did not guarantee compensation, establish causation, or determine which defendant—if any—was legally responsible.
What did the judge actually decide?
In May 2025, Judge Anne Conway allowed key claims to continue. At that stage, she declined to rule that Character.AI's chatbot outputs received first Amendment protection, according to the Associated Press report on the decision. The court also permitted claims involving Google's alleged development role to proceed.
That meant Garcia could continue litigating those theories; it did not establish that Google developed the relevant product or owed damages. This distinction is crucial. An early ruling that permits a claim to proceed is not a verdict. The plaintiff would still have needed to prove the required facts if the dispute had reached trial.
Why was there no immediate appeal?
Character Technologies and Google sought an interlocutory appeal—an appeal filed before the trial court has resolved the entire case. They wanted appellate review of issues involving the First Amendment and Google's potential liability. The district court denied both requests in July 2025.
Its order denying appeal certification said the proposed questions would not materially advance the litigation or required fact-specific review. That denial was procedural. It did not finally settle whether chatbot outputs qualify as protected speech or whether Google could be liable. Because certification was denied, the Eleventh Circuit did not decide those issues before the lawsuit ended.
What does the settlement establish about damages?
The parties announced a mediated settlement in principle in January 2026, and the federal action ended without a jury trial or merits verdict. The terms were not publicly disclosed, according to the Associated Press settlement report. The settlement therefore provides no public damages figure for comparison.
It also does not establish that the chatbot legally caused Setzer's death or create binding appellate precedent about AI-generated speech. Readers should be cautious with headlines describing the case as a courtroom victory or a decided liability standard. The settlement resolved this dispute without publicly answering the central factual and constitutional questions.
What can someone facing a similar situation do?
The Garcia complaint shows the types of claims one plaintiff pursued, but it does not create automatic eligibility for compensation. Any potential case would depend on its own communications, product use, injuries, defendants, and supporting records. A family evaluating a possible claim can preserve useful evidence before accounts or devices change: If someone may be in immediate danger, obtaining emergency help should take priority over collecting evidence.
- Save complete chatbot conversations rather than isolated excerpts.
- Keep screenshots with dates, times, usernames, and account details.
- Preserve relevant emails, product notices, device records, and subscription information.
- Create a factual timeline of product use, behavioral changes, medical events, and communications.
- Avoid editing original files, and give copies to a licensed attorney for review.