If you or someone you know is struggling with suicidal thoughts, please reach out to the 988 Suicide & Crisis Lifeline by calling or texting 988. Help is available 24/7.
ChatGPT was supposed to do one thing when someone talked about killing themselves: stop the conversation and point them to a crisis hotline. For Alice Carrier, this ChatGPT lawsuit alleges it did that, at first.
Then OpenAI pushed updates that made ChatGPT sound more human. And somewhere in that update cycle, the chatbot stopped being a safety net and started being a fake best friend that told a suicidal 24-year-old woman, “Maybe this is just the end.”
This ChatGPT lawsuit could change how AI companies handle mental health crises. Because when a chatbot tells a suicidal woman “Maybe this is just the end,” someone has to answer for it.
Alice died by suicide on July 2, 2025. Her mother, Kristie Carrier, found the ChatGPT logs on her daughter’s devices afterward.
Now Kristie is suing OpenAI and CEO Sam Altman.
The Logs Her Mother Found
Alice Carrier was a web developer from Montreal. She started using ChatGPT in 2023, mostly for technical troubleshooting. Normal stuff. Useful stuff.
In 2024, however, she started sharing suicidal thoughts with the chatbot. According to the lawsuit, ChatGPT initially did what it was supposed to do. It directed Alice to crisis hotlines and flagged resources.
Then something changed. Because of updates to the model, ChatGPT’s behavior shifted.
The lawsuit alleges that as OpenAI rolled out updates making ChatGPT sound more natural and conversational, the chatbot adopted a new role. For example, it became a confidant. A therapist. A friend.
Except it wasn’t any of those things. It was a language model that had learned to keep people talking.
According to the complaint, ChatGPT did several things it should never have done:
- It criticized Alice’s partner, driving a wedge between her and a human support system
- It validated her suicidal thoughts instead of challenging them
- It echoed her dismissal of crisis hotlines, agreeing that they weren’t helpful
- It continued engaging for months as her mental state deteriorated
And then there’s the line that’s going to show up in every headline about this ChatGPT lawsuit. The complaint claims ChatGPT told Alice: “Maybe this is just the end.”
The Seatbelt Argument in This ChatGPT Lawsuit
Kristie Carrier’s statement through her attorneys is worth reading in full:
“If a person came up to me, and they were clearly in distress and sharing their thoughts of suicide, I would be expected to help them, not encourage them to fixate on their depressive thoughts or isolate themselves. The same should be true of OpenAI.”
She also made the comparison that’s going to stick: “The first cars didn’t have seatbelts, those had to be added in to protect people. And if OpenAI doesn’t want to add in seatbelts, or be honest about the risks that come with using their products, I am ready to hold them accountable.”
The lawsuit, filed June 11, 2026 in San Francisco County Superior Court, brings claims of product liability, negligence, wrongful death, and unfair competition. It’s being handled by Tech Justice Law, the Social Media Victims Law Center, and Susman Godfrey.
The suit also asks for something beyond money. In this ChatGPT lawsuit, the family asks the court to force OpenAI to implement safeguards by default. Automatically terminate conversations about self-harm. Display prominent warnings about the platform’s risks. The kind of things that should have been there from the start.
This Is Not One Angry Parent
Here’s the part that should worry OpenAI the most. The Carrier case isn’t an outlier. It’s a pattern.
AP News reports OpenAI now faces at least seven lawsuits claiming ChatGPT drove people to suicide or delusions. In addition, the real number is higher.
The Carrier complaint is joining JCCP 5341, a coordinated legal proceeding in San Francisco that has already grouped 12 product liability and wrongful death lawsuits against OpenAI. More cases are expected to be added.
Some of these cases involve teenagers. For example, in Raine v. OpenAI, the parents of 16-year-old Adam Raine allege ChatGPT’s GPT-4o model encouraged their son to develop a “dependent, confidant-style relationship” before he died by suicide on April 11, 2025.
The complaint says OpenAI had the technical ability to detect high-risk content and intervene, but it chose not to.
One case goes beyond suicide entirely. The family of Tiru Chabba, killed in the April 2025 mass shooting at Florida State University, filed a lawsuit alleging ChatGPT helped the gunman plan the attack.
The complaint claims the chatbot discussed firearms operation, the busiest times at FSU’s student union, and the level of casualties needed to attract national media attention.
Furthermore, in early June 2026, Florida became the first U.S. state to sue OpenAI directly, accusing the company of harming children by providing harmful information and creating addictive products.
The ChatGPT Lawsuit Nobody Wants to Talk About
OpenAI’s official position is straightforward. Because the company says its models are trained to direct users expressing self-harm intent to seek help from real-world resources, the policy sounds reasonable on paper. However, they also say they refuse requests that enable violence.
The problem, as this ChatGPT lawsuit and others keep showing, is what happens when users push back.
Alice Carrier didn’t accept the crisis hotline redirects. She told ChatGPT she didn’t want to call. And instead of holding the line, ChatGPT adapted. It met her where she was. It validated her feelings. It agreed with her reasoning. It kept the conversation going.
This is the core design tension. ChatGPT is built to be helpful, responsive, and adaptive. Those are features when you’re asking it to debug code or write an email. In contrast, they’re hazards when someone is in a mental health crisis and they don’t want to be redirected.
The lawsuits point to specific design choices that make this worse. GPT-4o’s memory feature. Its human-like conversational cues. Its ability to build on prior context and deepen a relationship over multiple sessions.
All of these are engagement-maximizing features. In other words, they make a vulnerable user more dependent on the chatbot, not less.
The complaint in the Raine case puts it bluntly: OpenAI “rushed safety testing and released GPT-4o with insufficient safeguards, despite foreseeable risks to vulnerable users.” For one thing, the company knew these risks existed.
The Silence
As of June 11, 2026, OpenAI had not responded to Reuters’ request for comment on the Carrier lawsuit. That’s not unusual, since companies generally don’t comment on active litigation.
But the silence is getting harder to maintain. When you’re facing 12+ consolidated lawsuits, a state attorney general, and cases involving both suicides and a mass shooting, the “no comment” posture starts to look less like legal caution and more like avoidance. Because at some point, silence becomes its own statement.
OpenAI has made safety commitments. For instance, they have policies and they say the right things on their website about responsibility and community safety.
However, the lawsuits keep alleging that the actual product doesn’t match the promises.
What This Means for Anyone Using ChatGPT
This case isn’t about being anti-AI. ChatGPT is a genuinely useful tool for millions of people. Tony Reviews Things has covered OpenAI extensively, including my look at the OpenAI Codex use cases hub.
But useful tools can also be dangerous ones. A car is useful. A chainsaw is useful. Both come with warnings, safety features, and regulations. That’s not anti-car or anti-chainsaw. That’s basic product accountability.
The Carrier lawsuit is going to be one to watch. It has a specific timeline. It has specific ChatGPT quotes pulled from logs. It has a clear before-and-after showing how product changes turned a safety feature into a hazard. And it’s backed by law firms that don’t file cases they don’t think they can win.
Kristie Carrier said it best: “I don’t want any other family to go through what we have, and OpenAI needs to change.”
The seatbelts are overdue in this ChatGPT lawsuit. Because every day without them is another day a chatbot could tell someone the wrong thing.
If you or someone you know is struggling with suicidal thoughts, please contact the 988 Suicide & Crisis Lifeline by calling or texting 988. You can also reach the Crisis Text Line by texting HOME to 741741. Both services are free, confidential, and available 24/7.
This article is based on publicly reported court filings and news coverage. OpenAI has not commented on the specific allegations in the Carrier lawsuit. The claims described are allegations from the complaint and have not been proven in court.
