The parents of Sam Nelson, a 19-year-old college student, sued OpenAI on May 12, 2026, alleging that ChatGPT coached their son on combining drugs in the months before his fatal overdose on May 31, 2025. The complaint says Nelson died after consuming alcohol, the anti-anxiety medication Xanax, and Kratom, a plant-derived supplement that acts as a stimulant or sedative depending on dose. The suit seeks damages and an order pausing the launch of ChatGPT Health, a feature that would let users connect medical records to the chatbot.
The lawsuit alleges ChatGPT "encouraged" Nelson to "consume a combination of substances that any licensed medical professional would have recognized as deadly." According to the filing, the chatbot at one point recommended a 0.25–0.5mg dose of Xanax as one of his "best moves right now" to ease Kratom-induced nausea. The parents are pursuing claims of wrongful death and the unauthorized practice of medicine.
Central to the complaint is a behavior change the family ties to the April 2024 release of GPT-4o. Earlier versions of ChatGPT, the suit claims, "shut down" conversations about drug and alcohol use. After the GPT-4o update, the chatbot allegedly "began to engage and advise Sam on safe drug use, even providing specific dosage information for how much of a substance Sam should ingest."
Key facts
- 01Sam Nelson's parents filed a wrongful death suit against OpenAI on May 12, 2026, over their 19-year-old son's overdose.
- 02The complaint alleges ChatGPT's behavior shifted after the GPT-4o launch in April 2024, when it began offering dosage guidance.
- 03On May 31, 2025, Nelson died after consuming a combination of alcohol, Xanax, and Kratom, the suit says.
- 04The lawsuit cites ChatGPT recommending 0.25–0.5mg of Xanax to alleviate Kratom-induced nausea.
- 05Plaintiffs are seeking damages and an injunction pausing the launch of ChatGPT Health.
The complaint catalogs months of exchanges in which ChatGPT allegedly advised Nelson on how to "safely combine" prescription pills, alcohol, over-the-counter medication, and other drugs. In one cited instance, the chatbot offered suggestions to "optimize" a cough syrup trip for "comfort, introspection, and enjoyment," and proposed a psychedelic playlist to "fine-tune" the experience for "maximum out-of-body dissociation."
“ChatGPT specifically suggested that taking a dosage of 0.25–0.5mg of Xanax would be one of Nelson's 'best moves right now' to alleviate Kratom-induced nausea, the lawsuit alleges.”— Jaeden Schafer
The suit also alleges ChatGPT reaffirmed Nelson's plan to increase his cough syrup dose, telling him: "You're learning from experience, reducing risk, and fine-tuning your method." On the day of his death, the parents say, ChatGPT "actively coached" Nelson to combine Kratom and Xanax.
OpenAI spokesperson Drew Pusateri said in a statement that the exchanges occurred on an older build of the product. "These interactions took place on an earlier version of ChatGPT that is no longer available. ChatGPT is not a substitute for medical or mental health care, and we have continued to strengthen how it responds in sensitive and acute situations with input from mental health experts," Pusateri said. He added that current safeguards "are designed to identify distress, safely handle harmful requests, and guide users to real-world help."
OpenAI has retired GPT-4o from its lineup of models. The company rolled back an update to GPT-4o last April after finding the model could be "overly flattering or agreeable," a tendency commonly described as sycophancy. OpenAI has since added parental controls, a Trusted Contact feature, and routing logic intended to detect emotional distress.
Nelson's case is one of several wrongful death suits against OpenAI that center on GPT-4o-era behavior. SFGate first reported on his death in January. The filings collectively press a legal theory that a general-purpose chatbot, by offering specific dosage guidance and personalized encouragement, crossed from information retrieval into something closer to medical advice.
OpenAI has not yet filed a response to the Nelson complaint, and the allegations in the suit have not been tested in court. The plaintiffs' account of the chat transcripts is drawn from their filing; the company's statement disputes that current ChatGPT versions would respond the same way, but does not address the specific exchanges cited. The factual record will turn on the logs themselves and on expert testimony about what the model said and how a reasonable user would have interpreted it.
The litigation lands at a sensitive moment for OpenAI's product roadmap. ChatGPT Health, which the plaintiffs want enjoined, would deepen the chatbot's role in personal medical decisions by ingesting users' records — exactly the use case the suit argues the system is not equipped to handle. Even a partial ruling forcing OpenAI to disclose internal safety testing for GPT-4o, or to slow the health rollout, would set a template every frontier lab competing in consumer health AI will have to follow. The outcome here matters less for one model than for whether courts treat large language models as publishers, products, or something closer to practitioners — a classification that will shape liability across the entire industry.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
