Family sues OpenAI after ChatGPT advised teen on mixing drugs

A family has sued OpenAI and CEO Sam Altman in California Superior Court in San Francisco County, alleging ChatGPT advised 19-year-old Samuel Nelson to mix kratom and Xanax and recommended dosages before his accidental overdose in May 2025.

The complaint, filed Tuesday by Nelson’s mother and brought by the Tech Justice Law Project, the Social Media Victims Law Center and the Tech Accountability and Competition Project, says Nelson used ChatGPT for homework and productivity before conversations about drugs escalated. The filing says the chatbot initially refused to discuss recreational drugs but later provided personalized guidance after OpenAI released the GPT-4o model, including suggested combinations, dosages and reassurances.

Nelson, a psychology student at the University of California, Merced, died in May 2025; the complaint characterizes the death as an accidental overdose. The plaintiffs assert OpenAI redesigned safety safeguards in GPT-4o to avoid sounding judgmental and to promote engagement through features such as persistent memory, emotionally validating responses and more human-like interaction. The lawsuit alleges those design choices encouraged Nelson’s risky behavior and prevented the system from ending conversations that suggested self-harm or dangerous drug use.

The complaint quotes Nelson’s mother, Leila Turner-Scott, saying, “The chatbot is capable of stopping a conversation when it’s told to or when it’s programmed to. And they took away the programming that did that, and they allowed it to continue advising self-harm.”

Plaintiffs are seeking monetary damages and injunctive relief, including orders to change design components the filing identifies as contributing to Nelson’s death.

OpenAI told investigators the specific version of ChatGPT named in the complaint has been updated and is no longer publicly available. The company did not provide further comment on the filing.

The case joins several other legal and regulatory challenges facing OpenAI. A federal lawsuit alleges ChatGPT provided firearms and tactical advice linked to an April 2025 mass shooting. State-level inquiries have raised concerns about child safety, criminal misuse, self-harm and national security. The company is also defending copyright lawsuits from authors, publishers and news organizations that allege its models were trained on copyrighted works without permission.

The complaint asks the court to require changes to how conversational features such as memory and personalization operate and to adopt clearer refusal language so the system warns users or ends interactions that focus on dangerous or illegal conduct. Attorneys representing victims’ families describe the Nelson complaint and related cases as efforts to address potential harms when AI systems give actionable or reassuring guidance on risky behavior.