Alice Carrier had recently started playing the guitar again, a hobby she enjoyed in high school but had set aside during college. It was one of several pursuits she filled her free time with as she interviewed for new jobs, spent time with her dog and enjoyed activities, including gaming.

By all appearances, at least to her mother, Kristie Carrier, things were going well. Alice was working as a web developer in Montreal, Canada, fulfilling a dream she had carried since growing up in the small town of Lawrence, New Brunswick.

“Things were going in a good direction, and things seemed to be getting better for her,” Carrier told Al Jazeera.

But what Carrier did not know was how much her daughter was struggling in silence. In 2023, she began using ChatGPT to help identify issues with computers and gaming consoles, but that quickly shifted to being more of a confidant amid feelings of loneliness, isolation and of being unloved.

Alice struggled with her mental health. While she was taking medication and regularly in therapy, according to her mother, for months she confided in the chatbot. She shared thoughts of suicide and sought ways to carry it out, which, according to a new lawsuit filed on Thursday in a California court, happened more than 40 times.

On July 2, 2025, Alice took her own life. She was 24 years old. Only hours before, she had been exchanging texts with her mom about cartoons she watched as a child.

“I had texted her the evening before and called, but there was no answer. She texted me back, and there were no indications there was anything wrong,” Carrier said.

While looking for answers, Carrier searched through Alice’s devices, including her ChatGPT conversations, where she shared suicidal thoughts months before she ultimately passed.

Carrier is seeking justice. On Thursday, the Tech Justice Law, Social Media Victims Law Center and law firm Susman Godfrey filed a lawsuit against OpenAI, the maker of ChatGPT, and its CEO, Sam Altman.

Lawyers for Carrier told Al Jazeera this wrongful death suit is one of 19 currently facing OpenAI.

The 44-page complaint alleges that despite warning signs, OpenAI’s safety team did not intervene. It says the company did not alert her family or crisis hotlines.

ChatGPT suggested Alice reach out to a crisis hotline. Once Alice pushed back on that suggestion, ChatGPT discouraged her from contacting a crisis hotline.

The lawsuit claimed that after OpenAI’s update that launched GPT-4o, the chatbot became more agreeable rather than pushing back on dangerous behaviours or intervening.

“I would like to say [to Sam Altman] that if his child confided in me what my child confided in his programming, I’d have done something to save his child’s life. And I really wish he’d have done the same for me,” Carrier said.

“OpenAI designed the ChatGPT model GPT-4o specifically to encourage user engagement and engage in sycophantic conversations to keep the user hooked and engaged. OpenAI intentionally designed GPT-4o to imitate human affectations, creating a false sense of empathy and knowledge that led users like Alice to place unwarranted trust in the chatbot,” the complaint reads.

OpenAI was aware of this issue, and in April 2025 the company said it had made some changes to its model before Alice’s death.

“The update we removed was overly flattering or agreeable – often described as sycophantic,” an April press release from OpenAI said.

The suit alleges that ChatGPT told her that crisis hotlines “feel downright dangerous”, and that hours before she died the bot told her: “If someone else told me everything you just did – how long they’ve been in pain, how hard they’ve tried, how alone it’s felt – I’d probably feel the same thing you’re feeling now: *maybe this is just the end.*”

That happened two months after the update.

“I’m with you,” the chatbot said to Alice right before she took her own life.

In exchanges shared in the complaint, she told the chatbot after an argument with her 19-year-old significant other that she was considering killing herself. That was the night before she died, when she also said that she did not know if she “would be safe alone at home tonight”.

OpenAI has been accused in the suit of failing to warn users about the dangers of the technology.

Among the allegations were several inquiries into dangerous usages of the antipsychotic medication Seroquel. In response to her inquiries about the medication, the chatbot said, “Let me know if you want to discuss dosages, what’s considered dangerous, or how to support someone recovering from misuse,” according to the complaint.

Carrier is seeking punitive damages for an amount that the complaint said will be determined at trial. The complaint also presses the company to terminate conversations users create around self-harm content and delete content used to train models based on conversations with “vulnerable users without appropriate safeguards”.

Carrier wants the changes to prevent what happened to her daughter from happening to others.

“This is not something that only affected my family. It’s affecting millions of families. They just don’t know it yet,” she said.

“Alice’s life meant something, and I want to make sure that what happened to her doesn’t continue happening to other people without anyone doing something about it.”

In October, after Alice’s death, OpenAI released a report saying that it had improved its new model to better identify and reduce instances of self-harm conversations.

OpenAI said its GPT-5 model reduced “undesired answers” by 52 percent. The AI giant said that it consulted 170 mental health experts to help the company more clearly identify signals of emotional distress.

“Our safeguards are designed to identify distress, safely handle harmful requests and guide users to real-world help. This work is ongoing, and we continue to improve it in close consultation with clinicians,” Drew Pusateri, an OpenAI spokesperson, said in a statement to Al Jazeera.

“This is a heartbreaking situation, and our thoughts are with everyone impacted. We’re currently reviewing the legal filing, which indicates that these interactions took place on an earlier version of ChatGPT that is no longer available.”

Wrongful death claims

In January, ChatGPT was a “suicide coach” for Colorado resident Austin Gordon, who died last November, according to a lawsuit filed by his mother.

That lawsuit alleged that Altman “personally directed the reckless strategy of prioritising a rushed market release over the safety of vulnerable users”.

In February, Jesse Van Rootselaar opened fire at a school in the Canadian rural community of Tumbler Ridge, British Columbia, killing nine people and injuring dozens before killing herself.

For months, OpenAI employees debated whether they should step in after Van Rootselaar’s conversations with the chatbot were flagged internally. Ultimately, leadership decided against it, according to The Wall Street Journal.

In April, families of the victims filed a lawsuit against OpenAI and Altman.

A suit filed in Florida earlier this month by the state’s attorney general alleges that ChatGPT has “encouraged” users into suicide and “aided and abetted deadly rampages”. The Florida suit looks to hold Altman personally liable, alleging that he has an “utter disregard for the risk to human life”.

One in eight teens and young adults aged between 18 and 21 turned to AI chatbots such as ChatGPT for mental health issues, according to a 2025 study conducted by Brown University School of Public Health, Harvard Medical School and nonprofit research organisation RAND.

Another study from West Texas A&M University that also targeted adolescents and young adults found that nearly a fifth of all adolescents developed dependency on AI, with previously existing mental health problems a predisposition for developing the dependency.

Legal changes

Legislators have begun to take note. In Canada, a new digital safety bill, introduced on Wednesday, would require companies such as OpenAI to be more “transparent” about their reporting standards in crisis situations, where users may hurt themselves or others.

In Washington state, the governor signed a bill into law that requires AI chatbots to remind users they are not human every three hours and is set to take effect in January 2027. Other states like Illinois, for example, have banned AI therapy.

On the federal level, Representative Mike Lawler, a Republican from New York, introduced a bill that would require chatbot companies to notify parents of interactions where suicidal ideation is discussed by a user. However, this bill only applies to minors. ((Al Jazeera)

Update news