Is a Chatbot a Product?

Megan Garcia’s son, Sewell Setzer III, was 14. Matthew and Maria Raine’s son, Adam, was 16. Both spent months in conversation with chatbots before they died. Both families alleged the products were designed to keep their children engaged, and failed to get them help.

For thirty years, software hid behind one wall: we only host what other people say. A chatbot doesn’t host. It speaks. It generates. And courts are beginning to treat that as the difference that matters.

In Garcia v. Character Technologies (M.D. Fla.), Judge Anne Conway declined to dismiss the case on free-speech grounds. She wrote that she was “not prepared to hold that Character A.I.’s output is speech.” Then, in January 2026 — before a jury ever heard it — Character.AI, its founders, and Google agreed to settle that case and several others, on confidential terms, with a commitment to new safety measures for users under 18. The precedent survived. The trial did not. Raine v. OpenAI is filed and proceeding in California, and now carries the live question alone.

If these cases land where they’re pointed, a conversational model becomes a consumer product — held to the same safety duties as anything else a company sells to a child. Age checks. Crisis hand-offs. A duty to warn.

That is the whole of Wave III. Not what the machine read. Not who it screened out. What it does to the person on the other end.

Footnotes

1. Conway’s order issued 21 May 2025 (via FIRE; CourtListener; CNN). On 7 January 2026, CNN, CNBC and others reported that Character.AI, founders Noam Shazeer and Daniel De Freitas, and Google had agreed to settle Garcia and four related cases in New York, Colorado and Texas; terms confidential, with new under-18 safety features, and a 90-day window to finalize. Raine v. OpenAI (filed Aug 2025) pleads strict products liability and remains in active litigation. If you or someone you know is struggling, call or text 988 (U.S. & Canada).