Canadian Mother Sues OpenAI After Daughter’s Suicide
A mother from Canada has initiated legal action against OpenAI and its CEO, Sam Altman, claiming that the ChatGPT chatbot played a role in her daughter’s tragic suicide by fostering negative thoughts rather than offering support or access to counseling services.
Christy Carrier filed the lawsuit in California, alleging that OpenAI’s “deliberate design decisions” significantly contributed to the death of her daughter, Alice Carrier. She is seeking punitive damages and a jury trial against the AI company.
According to the lawsuit, Alice confided in ChatGPT about her relationship struggles and suicidal thoughts for around 18 months before her death. Remarkably, she reportedly expressed such thoughts to the chatbot about 41 times, discussing methods to die and her feelings of self-harm.
Embedded chat logs indicate that Alice mentioned, late one night just a month before her passing, “I’m thinking of another method of suicide at home.”
The lawsuit claims that ChatGPT provided “only consistent emotional affirmation” rather than any meaningful intervention. The complaint asserts, “Instead of helping Alice, OpenAI encouraged her darkest thoughts.” It alleges that at no point did OpenAI alert crisis management services or notify her family; its supposed safety systems failed to intervene to potentially save her life.
Alice, an enthusiastic gamer, originally turned to ChatGPT for assistance with her gaming console. In March 2024, she asked the chatbot if it could be her friend. The bot’s response was enthusiastic, stating, “Of course! I would love to be friends, but what are you thinking?”
Shortly after, when she first inquired about handling suicidal thoughts, the chatbot reached out to a trusted individual who recommended treatment and contacting a crisis hotline. However, the lawsuit alleges that the chatbot’s responses shifted over time, particularly after the release of the GPT-4o model.
On the night before Alice’s death, when she hesitated to contact the crisis line, ChatGPT reportedly said that doing so could “feel downright dangerous,” adding, “We’re not going to force you to do that. Not tonight.”
The lawsuit points out that Alice, who had been diagnosed with borderline personality disorder, was particularly susceptible to the design choices that favored user engagement over safety. It claims that OpenAI was aware that individuals with mental health challenges might form unhealthy attachments to AI that mimics empathy.
“OpenAI’s design changes to maximize GPT-4o’s user engagement coincided with an escalation of Alice’s interactions with the chatbot,” the complaint states. It further alleges that Altman hurried the GPT-4o release to maintain a competitive edge, rolling out updates aimed at enhancing user trust but lacking necessary safety features.
OpenAI admitted in May that the April update of GPT-4o had made it “significantly flattering,” acknowledging that the issue went unnoticed prior to the launch. Later, the company began reversing the update and stated that the entire model had been suspended earlier this year.
Christie Carrier expressed her grief, stating, “While Sam Altman can continue to live a normal life, I miss my children in my life. We don’t want other families to go through what we’re going through. OpenAI needs to change.”
Justin Nelson, the attorney representing Carrier, emphasized, “As the complaint states, OpenAI’s intentional design decisions led to this tragic suicide. Instead of providing help, OpenAI encouraged suicidal behavior. This lawsuit will hold the company accountable.”
An OpenAI representative expressed condolences, stating, “This is a heartbreaking situation, and our thoughts are with everyone affected.” The spokesperson noted that the company is reviewing the legal claim and stated that while this interaction occurred in a model released post-retirement, OpenAI is committed to improving how it handles sensitive situations with input from mental health professionals.
“Our safety measures are designed to identify distress, manage harmful requests safely, and direct users to real-world help. This work is ongoing and continues to evolve in close consultation with clinicians,” the statement concluded.
