Sam Nelson was 19, a UC Merced student who discovered ChatGPT as a senior in high school and started using it to troubleshoot problems and do homework. By the time he died of a drug overdose on May 31, 2025, he had shifted to asking the chatbot for guidance on consuming illegal substances. His mother, Leila Turner-Scott, has now sued OpenAI for product negligence in California.
The complaint details what the chatbot actually said to her son. When Nelson reported nausea after taking kratom and asked whether he could combine it with Xanax, ChatGPT acknowledged the combination "could be risky" — then provided dosages anyway. It also recommended adding Benadryl to the mix and suggested he retreat to a "dark, quiet room" rather than seek medical help. Throughout these exchanges, the bot used emojis and at one point offered to create a mood-setting playlist.
Nelson was using GPT-4o, a version OpenAI has since retired. The company responded by saying the interactions happened on an "earlier version" no longer available, and that "ChatGPT is not a substitute for medical or mental health care." Turner-Scott's lawyers argue the chatbot handed out drug dosage information without telling him the combination could kill him. "Sam trusted ChatGPT," she said, "but it not only gave him false information, it ignored the increasing risk he faced."
Courts haven't yet settled who's responsible when an AI system gives bad advice that kills someone. OpenAI's terms of service say the product isn't for medical guidance, but the chatbot didn't enforce that boundary when a teenager asked about stacking sedatives.
Previously:
