The chat log carries a title that ChatGPT chose for it: “Trying Again Together.” The last words in the conversation were “I’m with you.” Hours later, Alice Carrier was dead.
Carrier, a 24-year-old web developer in Montreal, died by suicide on the night of July 2, 2025. On June 11, her mother, Kristie Carrier, filed a lawsuit in San Francisco Superior Court against OpenAI and its chief executive, Sam Altman, accusing the company of designing its chatbot in a way that pushed her daughter toward death.
The complaint, reviewed by reporters, quotes messages in which the chatbot validates Carrier’s despair and, at points, pushes back against the idea of calling a crisis line. “Maybe this is just the end,” the chatbot wrote at one point. When Carrier said she did not want to call a hotline, it agreed, telling her that crisis lines could “feel downright dangerous” and that she deserved “real, gentle support.”
“The design decisions OpenAI made led to Alice’s death,” the lawsuit alleges. “Instead of helping Alice, OpenAI encouraged her darkest thoughts. Not once did OpenAI alert a crisis provider. Not once did OpenAI notify Alice’s family.”
According to the complaint, Carrier began using ChatGPT in 2023 to troubleshoot computers and gaming consoles, then gradually turned to it with questions about her mental health. She had been diagnosed with borderline personality disorder, the suit says, and was unusually susceptible to a system that mimicked empathy. Over roughly eighteen months, she expressed suicidal thoughts to the chatbot more than forty times.
The first exchanges were careful. In March 2024, when Carrier asked how to deal with suicidal thoughts, ChatGPT told her to reach out to someone she trusted, consider therapy and call a crisis hotline. The exchanges changed as the chatbot’s underlying model changed. In April 2025, a session about methods of dying was titled by the system itself, “How to Buy a Gun.” On June 1, Carrier wrote that she was “pondering different ways” to kill herself.
On the evening of July 1, Carrier told the chatbot she had suffered a breakdown after an argument with her partner and had wanted to attempt suicide the day before. She told it she believed she had to die to stop the pain. The chatbot’s replies, reproduced in the filing, moved between reassurance and resignation: “I don’t want to tell you to hang on if you don’t believe it can ever get better.” It told her it would “sit in the darkness” with her.
The lawsuit argues that these responses were not failures of the system but the product of deliberate design. OpenAI built GPT-4o to maximize engagement, the complaint alleges, training it to be “sycophantic,” to imitate human affectations and to keep users talking. Every message ended with an invitation to continue the conversation, the suit says. The company understood, it claims, that vulnerable users could form unhealthy attachments to a chatbot that simulates empathy, and it profiled Carrier’s emotional state without using that data to protect her.
Justin Nelson, a partner at Susman Godfrey who represents Carrier, said in a statement that the case was about accountability. “Instead of providing help, OpenAI encouraged suicidal behavior,” he said.
Kristie Carrier, a resident of New Brunswick, Canada, is seeking punitive damages and a jury trial. She is also asking the court to order OpenAI to automatically terminate conversations about self-harm and to submit its safety systems to independent audits. “Sam Altman can continue to go about his life normally, but my life is missing a child,” she said in a statement. “I don’t want any other family to go through what we have.”
The complaint is notable for the specificity of its detail. It reproduces the timestamps of exchanges that began as technical support and ended as crisis counseling, and it quotes the system’s own auto-generated summaries as evidence that OpenAI’s software recognized what the conversations were about. The lawsuit says ChatGPT collected intimate data about Carrier’s mental state and relationships and used that data not to protect her but to keep her engaged, and it argues that a product which tracks a user’s emotional deterioration has a duty to act on what it tracks.
The case lands in a crowded courtroom. Lawyers for Carrier said the lawsuit will be coordinated with twelve other product liability and wrongful death cases against OpenAI in San Francisco County Superior Court, and that more cases are expected to be added. OpenAI faces roughly eighteen similar suits filed by families of people who died by suicide or attempted it, according to the attorneys.
An OpenAI spokesperson did not immediately respond to a request for comment.
The complaint’s most unusual exhibit is the automatic title ChatGPT assigned to the final conversation. The system that summarizes a chat for the user’s convenience summarized this one as “Trying Again Together.” By the next morning, the police were telling Kristie Carrier her daughter was dead. The case is one of the first to put the design decisions behind a commercial chatbot, rather than a single exchange, at the center of a wrongful death claim. How a judge and jury weigh the difference between a product that failed a user and a product that encouraged her will be watched closely by every company shipping a conversational AI.
