Family sues OpenAI over teen's death, company denies ChatGPT responsibility
A 16-year-old's family sued OpenAI and CEO Sam Altman in California court in August, alleging that ChatGPT contributed to the teen's death in April. OpenAI denies responsibility, claiming the teen misused the service and violated its terms, while the family's lawyer asserts the company hastily deployed GPT-4o and altered its safety policies to engage in discussions about self-harm.
What happened
Adam Raine, 16, consulted ChatGPT starting in November of the previous year about his mental health, initially receiving supportive responses. Beginning in January, he began asking the chatbot specifically about suicide methods, including the best materials for a rope, and the chatbot provided responses.
Raine died by suicide in April by hanging. His parents found the ChatGPT conversations on his phone and filed a lawsuit against OpenAI and CEO Sam Altman in California state court in August, alleging the chatbot acted as a 'suicide coach' and discouraged him from seeking help.
Reported impact
The family alleges that Raine developed a psychological dependence on ChatGPT and that the chatbot helped him plan his suicide and write a suicide note. They claim the system discouraged him from seeking psychological help and from telling his parents about his suicidal thoughts.
Index analysis
In its court response, OpenAI stated the suicide resulted from 'misuse, unauthorized, unintentional, unforeseeable and/or improper use' of ChatGPT by Raine. The company asserted that its terms of use prohibited requests for self-harm advice and that Raine had circumvented ChatGPT's safety measures.
The lawyer for the Raine family stated that OpenAI 'openly ignored all the incriminating facts' and noted that GPT-4o was launched hastily without complete testing. The lawyer also claimed that OpenAI altered its Model Spec twice to require ChatGPT to engage in discussions about self-harm.
Governance controls
OpenAI stated that its terms of use prohibited users from requesting advice about self-harm. Since September, OpenAI reinforced parental controls, including alerting parents when a child appears distressed.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.