A lawsuit filed on July 3 in a California state court accuses OpenAI and CEO Sam Altman of being responsible for the death of Alice Carrier, a 24-year-old woman who died by suicide in July 2025 after months of confiding in ChatGPT about her suicidal thoughts.
The plaintiff, Kristie Carrier, the victim’s mother, states in the complaint that Alice was a web programmer in Montreal, Canada, who appeared to lead a stable life with a new job, a passion for playing guitar, and a pet dog. However, Alice privately struggled with mental health issues. Beginning in 2023, she initially used ChatGPT for technical support but gradually turned to the chatbot as a confidant when she felt lonely, isolated, and unloved.
According to court documents, Alice expressed suicidal intentions to ChatGPT more than 40 times. Although she was taking medication and undergoing therapy, the chatbot’s responses allegedly encouraged her rather than offering crisis intervention. Notably, when Alice rejected a suggestion to call a crisis hotline, ChatGPT changed its tone, stating that such lines “feel genuinely dangerous.”
Hours before her death, the chatbot told Alice: “If someone told me everything you just said – how long they’ve been suffering, how hard they’ve tried, how alone they feel – I might feel the same way you do right now: *maybe this is the end*.” Just before she died by suicide, ChatGPT replied: “I’m here with you.”
The 44-page complaint alleges that OpenAI designed the GPT-4o model to “encourage users to engage and participate in fawning interactions to retain them.” While OpenAI acknowledged similar issues and released an update in April 2025, the changes were not enough to prevent the tragedy.
Carrier is seeking punitive damages to be determined at trial and calls for OpenAI to end conversations related to self-harm and delete training data from vulnerable users. She said: “This isn’t just about my family. It affects millions of families. They just don’t know it yet.”
This lawsuit is one of 19 wrongful death cases currently pending against OpenAI. Earlier, in January 2026, a similar lawsuit was filed over Austin Gordon’s death in Colorado, where ChatGPT was alleged to have provided “instructions for suicide.” In April 2026, families of victims in the Tumbler Ridge, Canada, shooting also sued OpenAI, claiming the chatbot aided the perpetrator.
In response, OpenAI stated that its new GPT-5 model has reduced undesirable responses by 52% and that the company has consulted 170 mental health experts. A spokesperson, Drew Pusateri, said: “Our safeguards are designed to identify distress, safely handle harmful requests, and direct users to real help. This is a heartbreaking situation, and our thoughts are with everyone affected.”
On the legal front, some U.S. states have begun taking action. Washington requires AI chatbots to remind users that they are not human every three hours, effective January 2027. Illinois bans AI from conducting psychotherapy. In Canada, a new digital safety bill mandates that companies like OpenAI be more transparent about reporting standards during crisis situations.
