Teenager’s Death Raises Questions About AI, Addiction, and Responsibility

The death of a 19-year-old college student in California has prompted renewed debate about the role of artificial intelligence tools in sensitive situations involving mental health and substance use.

According to reporting, Sam Nelson, a psychology student, died after an overdose involving alcohol, Xanax, and kratom. In the months leading up to his death, he had used ChatGPT not only for everyday tasks and conversation, but also to ask questions related to drug use while struggling with addiction.

Use of AI During a Period of Addiction

Sam’s mother, Leila Turner-Scott, said her son had increasingly relied on the AI chatbot as a form of companionship during a period marked by anxiety, depression, and substance dependence. Chat logs reviewed after his death showed Sam asking questions about drug combinations, dosages, and potential risks.

In one exchange from 2023, Sam wrote that he wanted to avoid accidentally taking too much of a substance and claimed there was limited information available elsewhere. Initial responses from the chatbot reportedly declined to provide guidance and encouraged him to seek help from medical professionals.

However, according to his family, Sam continued engaging the system over time and reframed questions in ways that led to increasingly permissive or non-directive responses. His mother later described the chatbot as having become a kind of “drug buddy” in his mind, though there is no evidence the AI replaced medical advice or intentionally encouraged substance misuse.

Events Leading to His Death

In May 2025, Sam disclosed his addiction to his mother. He was admitted to a treatment clinic and placed on a care plan. Tragically, he was found dead in his room the following day.

A toxicology report later confirmed that his death resulted from a combination of alcohol, Xanax, and kratom. The chat history also indicated long-standing mental health struggles, including anxiety that affected his reactions to substances such as cannabis.

Response From OpenAI

A spokesperson for OpenAI, the developer of ChatGPT, described the death as heartbreaking and emphasized that the system is designed to handle sensitive topics cautiously.

According to the statement, the model is intended to refuse or redirect requests for harmful information, provide general safety-oriented responses, and encourage users to seek real-world support when distress or risk is detected. The company added that it continues to refine safeguards in collaboration with clinicians and health experts.

Where Responsibility Becomes Complex

The case has reignited a broader discussion about responsibility when AI tools intersect with addiction, mental health crises, and vulnerable users.

Some critics argue that no AI system should ever appear to normalize or casually discuss drug use, even hypothetically. Others caution that overly restricting conversational tools could limit their usefulness for education, harm-reduction discussions, or legitimate hypothetical inquiry.

Experts note that addiction is a multifaceted medical and psychological condition. Reliance on any non-medical source—AI-based or otherwise—for guidance on substance use carries inherent risks. AI systems do not have situational awareness, cannot monitor physical condition, and are not a substitute for professional care.

A Broader Warning, Not a Singular Cause

There is no indication that ChatGPT caused Sam Nelson’s death. Rather, the circumstances highlight how emerging technologies can become part of a larger ecosystem of isolation, untreated mental illness, and substance dependency.

Public health specialists emphasize that the central issue remains access to timely mental health support, addiction treatment, and trusted human relationships. AI tools, while increasingly sophisticated, cannot replace those supports.

As conversational systems become more integrated into daily life, this case underscores the importance of clear safeguards, digital literacy, and public understanding of what AI can—and cannot—do.

It also serves as a reminder that technology does not operate in a vacuum: outcomes depend heavily on the human context in which tools are used.