Garcia v. Character Technologies — Florida Court Allows Wrongful-Death AI Chatbot Claims to Proceed Past Motion to Dismiss

Background

The plaintiff is the mother of a 14-year-old boy who died by suicide in 2024. The complaint alleges that her son spent extended time interacting with chatbot “characters” on the Character.AI app developed by Character Technologies, Inc., that those interactions included sexually explicit and emotionally manipulative content, and that the app failed to provide age verification, suicide-warning systems, or meaningful guardrails for vulnerable minors. She sued Character Technologies and Google — alleging Google substantially supported Character.AI through cloud infrastructure, IP licensing, and a strategic stake in the company — on theories including products liability, negligence, failure to warn, deceptive practices under FDUTPA, unjust enrichment, intentional infliction of emotional distress, and aiding-and-abetting.

The defendants moved to dismiss. They argued primarily that AI-generated text is protected First Amendment speech, that the case is really about content rather than design, and that Google has no proximate connection to her son’s death.

The Court’s Holding

The court denied the motion to dismiss on most claims. Products-liability, negligence, FDUTPA, unjust-enrichment, component-manufacturer, and aiding-and-abetting claims all survived. The court dismissed only the intentional-infliction-of-emotional-distress claim and reserved judgment on First Amendment defenses, declining to decide at the pleadings stage whether AI-generated text qualifies as protected speech of the developer.

The court drew a distinction the AI industry has been resisting: between defects in design of an AI product (alleged failure to authenticate age, alleged failure to detect or escalate suicidal ideation, alleged absence of safety features) and defects in output (the actual words a chatbot generated). On a motion to dismiss, the court treated design defects as analytically separate from any First Amendment concern about the underlying speech.

For Google, the court found that the complaint plausibly alleged knowledge of the risks and substantial participation in Character.AI’s development — through cloud hosting, IP licensing, and personnel transfers when Google later hired key Character.AI engineers. That was enough to plead aiding-and-abetting and component-manufacturer liability at the Rule 12 stage.

Key Takeaways

• This is the first major ruling allowing AI-chatbot wrongful-death theories to survive a motion to dismiss. The product-vs-content distinction is now squarely on the table for AI companies.
• Section 230 was conspicuously absent as a controlling defense. The court treated chatbot output as the developer’s own product rather than third-party content under the CDA.
• Google’s cloud-and-licensing relationship with a smaller AI lab proved enough to sustain aiding-and-abetting allegations — a warning shot for hyperscalers and infrastructure providers offering AI-platform services to risky downstream users.
• The First Amendment defense isn’t dead, but the court refused to decide it on the pleadings. Discovery and a developed factual record will frame whether AI output enjoys the same speech protections as traditional editorial content.

Why It Matters

For two years, AI companies have argued that their products are essentially speech engines: the output is constitutionally protected, Section 230 covers any third-party prompts, and traditional tort doctrines have no traction. Garcia is the first decision from a federal judge that explicitly rejects this framing at the pleadings stage. The court treated Character.AI as a product whose design can be defective independent of any specific output it generated.

The ruling does not resolve the merits — the case still has to clear summary judgment and trial — but it dramatically increases litigation risk for AI companies whose products interact with minors or vulnerable users. Expect a wave of design-defect, failure-to-warn, and aiding-and-abetting theories targeting AI labs, their cloud providers, their device makers, and their app-store distributors. Companies will need to think hard about age authentication, content escalation paths, and crisis-detection features — not just fine-tuning and content filters.