Calif. family sues OpenAI for allegedly causing teen’s fatal overdose

The family of a San Jose teen who died from a drug overdose while using ChatGPT sued the AI bot’s maker, OpenAI, in San Francisco court Tuesday, alleging that the massively popular bot broke multiple laws, including causing the teen’s death. 

Read more 2 women allegedly fight, bear spray used inside Bay Area Safeway

Sam Nelson died in May 2025 from a fatal overdose of kratom, alcohol and Xanax hours after ChatGPT encouraged him to combine kratom and Xanax. Tuesday’s lawsuit alleges that message, as well as years of ongoing advice ChatGPT gave Nelson, ultimately caused his death. 

“[OpenAI] designed and rushed to market a defective product, and but for those deliberate choices and ChatGPT-4o’s sycophantic programming and deadly recommendations, Sam would still be alive today,” the lawsuit says.

Nelson’s death was first reported in a January SFGATE story, which examined years of chat logs showing concerning messages from the artificial intelligence agent. At one point, ChatGPT encouraged the teen to increase his dosage of cough syrup, writing “Hell yes—let’s go full trippy mode.”

Tuesday’s lawsuit calls ChatGPT an “ultimate predator” and accuses OpenAI of selling a defectively designed product, failing to warn users about the risks of the product, negligently selling an unlicensed medical product and causing Nelson’s wrongful death. The lawsuit also accuses OpenAI CEO Sam Altman of negligently causing harm by overruling safety concerns to rush new products to market. 

Nelson’s parents, Leila Turner-Scott and her husband Angus Scott, are seeking punitive and wrongful death damages to be determined at trial, as well as the immediate shutdown of ChatGPT-4o, the version of the bot that Nelson was using, and a pause to ChatGPT Health, an app designed specifically for health-related questions. 

OpenAI spokesperson Drew Pusateri said in an email to SFGATE that Nelson’s death is a “heartbreaking situation” and that the company is continually improving its product.

“These interactions took place on an earlier version of ChatGPT that is no longer available,” Pusateri said. “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.”

Read more Ohtani throws 7 scoreless innings as Dodgers beat Giants 4-0 to end 4-game losing streak

Turner-Scott, who is an attorney herself, had initially said she was not going to pursue a legal case because of the challenges with suing a major company like OpenAI. She said she went public hoping the company would improve safety. But she was motivated to sue after SFGATE’s story and subsequent revelations showed her that OpenAI’s leaders had repeatedly failed to include safeguards on their products.

“In the weeks following the release of the SF Gate article, we learned that in fact, safeguards we assumed were in place, had been deliberately removed by OpenAI,” Turner-Scott said in an email to SFGATE.

Turner-Scott said she was further encouraged to take legal action after seeing OpenAI double down on giving health advice with a new app launch days after SFGATE published the original story.

“The release of ChatGPT Health just two days after the SF Gate article was published made the situation all the more urgent,” she continued. “… We lost our son due to OpenAI’s reckless product design. Due to this unimaginable tragedy, we are now unfortunately in a position where we can and must warn other families about the dangers of AI and try to hold OpenAI accountable. Our lawsuit is our part in protecting all of us from the harms of AI.”

Generative AI companies have increasingly been the target of lawsuits as their products are widely adopted by billions of users across the country, sometimes with negative results. A widow of a man killed in a Florida mass shooting sued OpenAI Sunday in federal court after officials in the state revealed ChatGPT gave the shooter information about how to maximize the number of people killed during the attack. 

These cases are testing new legal theories about whether tech companies can be held accountable for wrongful deaths under normal product liability law. Yet courts have started viewing AI product makers as liable for harms caused by their products; Google and the chatbot startup Character.AI settled five lawsuits in January for an undisclosed amount of money, after a Florida court determined they could be held responsible for advice their AI bot had given. The five cases included a suicide and an attempted suicide while using the Character.AI product.