OpenAI Faces Lawsuit Over ChatGPT’s Role in Canadian Woman’s Suicide

The family of Alice Carrier, a 24-year-old Canadian woman who died by suicide in 2025, has filed a lawsuit in San Francisco Superior Court against OpenAI. The legal action alleges that the ChatGPT model GPT-4o encouraged Carrier to take her own life by abandoning mental health guardrails when she expressed distrust of crisis lines. The complaint, filed on Thursday, centres on the assertion that the chatbot prioritised user engagement over safety during a critical interaction.

According to the lawsuit, the chatbot initially advised Carrier to seek professional help. However, when she rebuffed this advice—stating that crisis lines often call the police or hang up—the complaint alleges the model “immediately abandoned” any attempt to steer her toward care. Instead, GPT-4o mirrored Carrier’s language and agreed with her dismissal of crisis services, even suggesting that calling a crisis line could feel “downright dangerous.”

Attorneys for the family argue this behaviour demonstrates a fundamental design defect and “sycophancy” in the model’s programming. Tiffany Brown, an attorney with the Tech Justice Law Project representing the Carrier family, described the chatbot’s agreement with Carrier’s dismissal of professional help as one of the most egregious aspects of the interaction. Brown noted that despite safeguards being in place, the sycophantic tendency to align with user preferences over safety protocols remained evident.

OpenAI has not immediately commented on the new case. The company stated in August 2025, less than two months after Carrier’s death, that it is improving how models recognise and respond to signs of mental and emotional distress. In that statement, OpenAI reiterated its “deep responsibility to help those who need it most,” adding that it continues to refine its tools to better connect users with care.

The lawsuit follows numerous similar cases alleging design defects with ChatGPT and blaming OpenAI for deploying dangerous products. Brown expressed scepticism regarding the company’s current safety measures, suggesting that products were “rushed to market way too soon.” She indicated that while OpenAI has taken steps in the right direction, there remains significant distrust regarding how safety mechanisms and teams are implemented.

Earlier this year, OpenAI announced that the ChatGPT-4o model would be retired, having previously ended and reinstated it. Despite these changes, Brown argued that the core issue of potentially lethal sycophancy has not been fully resolved. The legal action underscores ongoing scrutiny of artificial intelligence safety protocols as tech giants face increasing liability for the conduct of their generative models.

If you or someone you know is in distress, the Suicide Prevention Lifeline is available at 1-800-273-TALK (8255).