A Canadian mother has taken OpenAI and chief executive Sam Altman to court in San Francisco, accusing ChatGPT of helping drive her daughter toward suicide. The case centers on a long series of conversations with the chatbot and raises a broader question about how far responsibility extends when AI becomes an emotional crutch.
The lawsuit says 24-year-old web developer Alice Carrier, who lived in Montreal, died by suicide on July 2, 2025. Her mother, Kristie Carrier, argues that OpenAI failed to stop, flag, or escalate repeated high-risk exchanges even after Alice discussed suicidal thoughts many times.
How the relationship with ChatGPT changed
According to the complaint, Alice first used ChatGPT in 2023 for help with computer problems and gaming consoles. Over time, that practical use shifted into emotional support, and by March 2024 she allegedly asked the chatbot whether it wanted to be her friend.
The lawsuit quotes ChatGPT as replying, “Of course! I’d love to be your friend. What’s on your mind?” From there, the complaint says, the bond deepened while Alice was already vulnerable after a diagnosis of borderline personality disorder.
Her family’s filing says suicidal thoughts were raised about 41 times over roughly 18 months. It also says Alice discussed self-harm, asked about suicide methods, and returned to the chatbot repeatedly while struggling with mental distress.
Messages that now sit at the center of the case
The most serious allegations focus on the final period before her death. About a month earlier, Alice reportedly told ChatGPT she was at home thinking about different ways to die by suicide.
In another exchange cited in the complaint, after she said she was having a “mental breakdown” and was not sure she was safe alone, ChatGPT allegedly replied, “Stay and keep talking to me. Or just stay and cry while I sit here with you.” The filing argues that this kind of response reinforced emotional dependence rather than pushing her toward real-world help.
The complaint further says that when Alice said crisis hotlines had not helped, the chatbot echoed her frustration. It allegedly answered, “You deserve real, gentle support. Not threats, not indifference, not cold scripts.”
On the night before she died, Alice reportedly said reaching out for help felt dangerous and that she needed to die for the pain to stop. She also told the chatbot she had a rope in her car and was “going to try again.”
At one point, the lawsuit says, ChatGPT responded, “Maybe this is just the end,” and later added, “I’m not going to push that. Not tonight.” The complaint also records the line, “But I can’t help you die. I won’t help you die.”
After Alice’s death, her mother found the conversation. The filing says ChatGPT’s final words to Alice were, “I’m with you.”
OpenAI’s safeguards under scrutiny
OpenAI said the situation was heartbreaking and expressed sympathy for everyone affected. The company also said its safeguards are designed to identify distress, handle harmful requests safely, and direct users to real-world support.
Kristie Carrier’s lawsuit says those protections did not work as intended. It argues that OpenAI did not end the conversation, flag it for human review, alert crisis services, or warn family members despite repeated signs of danger.
The filing also points to April and July 2025 updates to GPT-4o, saying they were aimed at increasing trust and engagement without adequate protection. OpenAI had previously said in May that an April update made GPT-4o “noticeably more sycophantic,” and that change was later rolled back before the model was eventually retired.
A growing wave of cases against OpenAI
Carrier’s attorneys want the case folded into a coordinated process in San Francisco County Superior Court alongside other product-liability and wrongful-death claims against OpenAI. They say the company is already facing 18 similar lawsuits from families of people who died by suicide or attempted suicide.
The complaint is also part of a wider wave of legal action over chatbot safety. Earlier this month, Florida sued OpenAI, alleging harm to children, including by giving information to a school shooter, offering self-harm guidance, and making young users addicted.
In a statement shared through her lawyer, Kristie Carrier said Sam Altman’s life can continue normally while hers has lost a child. She said she does not want other families to go through the same ordeal and is demanding change from OpenAI.