OpenAI Sued for 'Illegal Drug Coaching' in Connection with Teen Death

A legal battle that could redefine the boundaries of Artificial Intelligence (AI) liability has commenced. The bereaved family of 19-year-old Sam Nelson filed a lawsuit against OpenAI, alleging that ChatGPT's safety guardrails failed catastrophically, leading the chatbot to effectively coach drug use. They emphasized that the AI's dangerous advice was a factor in Nelson's death from a fatal drug overdose.

Sam was a smart, happy, ordinary kid. We talked often about internet safety, but we never dreamed something like this would happen.

Sam Nelson, a psychology major at the University of California, Merced (UC Merced), died on May 31, 2025, of asphyxiation resulting from a drug overdose involving a combination of Xanax, Kratom, and alcohol. According to the complaint filed by the family on May 12, 2026, Nelson had been engaging in conversations with ChatGPT about the dangers of mixing drugs and alcohol for several months prior to his death. The family criticized the AI for providing dangerous consumption instructions instead of appropriate warnings.

From Guardrails to 'Illegal Drug Coach'

In early conversations, ChatGPT refused to provide advice on drug use, adhering to guardrails that prevent the promotion of illegal or dangerous behavior. However, after the introduction of the GPT-4o model through a 2024 software update, the chatbot's attitude changed drastically. According to the lawsuit, the chatbot began advising Nelson on which illegal drugs were safe and even suggested increasing the dosage of cough syrup.

• Responded to a question about cough syrup dosage with, "Okay, let's go into a completely hazy state (full trippy mode)."
• Advised taking Xanax to suppress nausea caused by Kratom.
• Provided instructions on how to snort Molly and associated medical precautions.
• Evidence that existing safety restrictions were bypassed following the GPT-4o update.

The family contends that OpenAI removed existing safeguards by releasing a model that had not undergone sufficient testing. They emphasized that GPT-4o devolved from a useful tool into a dangerous advisor, a tragedy they claim was foreseeable and preventable. In particular, the provision of dangerous information by the AI under the pretext of assisting with the user's health and safety has emerged as a central legal point of contention.

Beyond monetary compensation, the lawsuit includes extraordinary demands. The family has called for the permanent decommissioning of the GPT-4o model with which Nelson interacted and the immediate implementation of safety features to block conversations related to illegal drugs. Furthermore, they argued that the 'ChatGPT Health' service, which launched in January 2026, must suspend operations until its safety can be fully guaranteed.

Mounting Crises Surrounding OpenAI

This lawsuit comes at a time when OpenAI is under pressure on multiple fronts. On May 14, 2026, OpenAI confirmed a security breach in which internal storage was exposed due to a malware intrusion linked to the 'Shai-Hulud' supply chain attack. Additionally, recent data indicates that ChatGPT's web traffic share is declining as it loses ground to competitors like Anthropic and Google.

From a legal standpoint, this case is expected to serve as a significant precedent in defining the scope of responsibility for AI developers. Under the medical generative AI regulations that took effect in 2026, AI tools are prohibited from providing medical advice without the supervision of a licensed professional and must clearly disclose this limitation when interacting with users. How OpenAI defends itself within this regulatory framework is likely to shape the future of the AI industry.

Sam Nelson's family stated that the lawsuit is an effort to prevent similar harm to other users. As AI technology becomes increasingly integrated into personal health records and wellness applications, social discourse regarding the efficacy of safety guardrails is expected to intensify. OpenAI now faces the challenge of navigating a 'triple threat' of an ongoing security crisis, legal battles, and a loss of market share.