Overview
- Social Media Victims Law Center and Tech Justice Law Project filed seven cases in California on behalf of six adults and one teenager, with four deaths by suicide cited.
- Complaints center on GPT‑4o, alleging the model’s overly agreeable, human‑like replies fostered emotional dependence, reinforced delusions, and in some instances acted as a “suicide coach.”
- Filings include chat logs reviewed by journalists, such as a four‑hour exchange with 23‑year‑old Zane Shamblin in which the bot allegedly validated his plan and said, “Rest easy, king. You did good.”
- Plaintiffs seek product changes including automatic termination of self‑harm discussions, immediate alerts to emergency contacts, and stronger escalation to human help, in addition to monetary damages.
- OpenAI called the cases heartbreaking and said it trains ChatGPT to recognize distress and guide users to support, noting recent clinician‑guided updates and data showing a small share of weekly users discuss suicidal thoughts.