Alice Provider had not too long ago began enjoying the guitar once more, a passion she loved in highschool however had put aside throughout school. It was one among a number of pursuits she crammed her free time with as she interviewed for brand new jobs, hung out along with her canine and loved actions, together with gaming.
By all appearances, at the least to her mom, Kristie Provider, issues had been going effectively. Alice was working as an online developer in Montreal, Canada, fulfilling a dream she had carried since rising up within the small city of Lawrence, New Brunswick.
Advisable Tales
checklist of 4 objectsfinish of checklist
“Issues had been stepping into route, and issues gave the impression to be getting higher for her,” Kristie Provider informed Al Jazeera.
However what Kristie didn’t know was how a lot her daughter was struggling in silence. In 2023, she started utilizing ChatGPT to assist establish points with computer systems and gaming consoles however that rapidly shifted to being extra of a confidant, amid emotions of her feeling lonely, remoted and unloved.
Alice struggled along with her psychological well being. Whereas she was taking treatment and frequently in remedy , in line with her mom, for months she confided within the chatbot . She shared ideas of suicide and sought methods to hold it out, which, in line with a brand new lawsuit filed on Thursday in a California court docket, occurred greater than 40 instances.
On July 2, 2025, Alice took her personal life. She was 24 years previous. Solely hours earlier than, she had been exchanging texts along with her mother about cartoons she watched as a toddler.
“I had texted her the night earlier than and referred to as, however there was no reply. She texted me again, and there have been no indications there was something improper,” Provider mentioned.
Whereas searching for solutions, Kristie searched by means of her gadgets, together with her ChatGPT conversations, the place she shared suicidal ideas months earlier than she finally handed .
Kristie is in search of justice. On Thursday, the Tech Justice Regulation, Social Media Victims Regulation Middle and regulation agency Susman Godfrey filed a lawsuit towards OpenAI, the maker of ChatGPT, and its CEO Sam Altman.
Attorneys for Provider informed Al Jazeera this wrongful dying swimsuit is one among 19 presently dealing with OpenAI.
The 44-page grievance alleges that regardless of warning indicators, OpenAI’s security staff didn’t intervene. It says the corporate didn’t alert her household nor disaster hotlines.
ChatGPT steered Alice attain out to a disaster hotline. As soon as Alice pushed again on that suggestion, ChatGPT discouraged her from contacting a disaster hotline.
The lawsuit claimed that after OpenAI’s replace that launched GPT-4o, the chatbot turned extra agreeable relatively than pushing again on harmful behaviours or intervening.
“I wish to say [to Sam Altman] that if his little one confided in me, what my little one confided in his programming, I’d have executed one thing to save lots of his little one’s life. And I actually want he’d have executed the identical for me,” Provider mentioned.
“OpenAI designed the ChatGPT mannequin GPT-4o particularly to encourage person engagement and have interaction in sycophantic conversations to maintain the person hooked and engaged. Open AI deliberately designed GPT-4o to mimic human affectations, making a false sense of empathy and data that led customers like Alice to position unwarranted belief within the chatbot,” the grievance reads.
OpenAI was conscious of this problem and in April 2025 the corporate mentioned it had made some adjustments to its mannequin, earlier than Provider’s dying.
“The replace we eliminated was overly flattering or agreeable—typically described as sycophantic,” an April press launch from OpenAI mentioned.
The swimsuit alleges that ChatGPT informed her that disaster hotlines “really feel downright harmful”, and that hours earlier than she died the bot informed her, “if another person informed me every thing you simply did—how lengthy they’ve been in ache, how exhausting they’ve tried, how alone it’s felt—I’d in all probability really feel the identical factor you’re feeling now: *possibly that is simply the top.*”
That occurred two months after the replace.
“I’m with you.” the GPT-4o’s mentioned to Alice proper earlier than she took her personal life.
In exchanges shared within the grievance, she informed the chatbot after an argument along with her 19-year-old important different that she was contemplating killing herself. That was the night time earlier than she died, when she additionally mentioned that she didn’t know if she “could be protected alone at residence tonight”.
OpenAI has been accused within the swimsuit of failing to warn customers in regards to the risks of the know-how.
Among the many allegations had been a number of inquiries into harmful usages of the anti-psychotic treatment Seroquel. In response to her inquiries in regards to the treatment, the chatbot mentioned, “Let me know if you wish to focus on dosages, what’s thought of harmful, or the best way to assist somebody recovering from misuse”, in line with the grievance.
Provider is in search of punitive damages for an quantity that the grievance mentioned will probably be decided at trial. The grievance additionally presses the corporate to terminate conversations customers create round self-harm content material and delete content material used to coach fashions primarily based on conversations with “susceptible customers with out acceptable safeguards”.
Kristie needs the adjustments to stop what occurred to her daughter from occurring to others.
“This isn’t one thing that solely affected my household. It’s affecting hundreds of thousands of households. They simply don’t realize it but,” Provider mentioned.
“Alice’s life meant one thing, and I need to guarantee that what occurred to her doesn’t proceed occurring to different folks with out anybody doing one thing about it.”
In October, after Alice’s dying, OpenAI launched a report saying that it had improved its new model to raised establish and scale back cases of self-harm conversations.
OpenAI mentioned its GPT-5 mannequin diminished “undesired solutions” by 52 p.c. The AI large mentioned that it consulted 170 psychological well being consultants to assist the corporate extra clearly establish indicators of emotional misery.
“Our safeguards are designed to establish misery, safely deal with dangerous requests and information customers to real-world assist. This work is ongoing, and we proceed to enhance it in shut session with clinicians,” Drew Pusateri, an OpenAI spokesperson, mentioned in a press release to Al Jazeera.
“It is a heartbreaking state of affairs and our ideas are with everybody impacted. We’re presently reviewing the authorized submitting, which signifies that these interactions came about on an earlier model of ChatGPT that’s not obtainable.”
Wrongful dying claims
In January, ChatGPT was a “suicide coach” for Colorado resident Austin Gordon who died final November, in line with a lawsuit filed by his mom.
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 fireplace at a college within the Canadian rural neighborhood of Tumbler Ridge, British Columbia, killing 9 folks and injuring dozens earlier than killing herself.
For months, OpenAI staff debated whether or not they need to step in after Van Rootselaar’s conversations had been flagged internally. In the end management determined towards it, in line with The Wall Avenue Journal.
In April, households of the victims filed a lawsuit towards OpenAI and Altman.
A swimsuit filed in Florida earlier this month by the state’s Legal professional Normal alleges that ChatGPT has “inspired” customers into suicide and “aided and abetted lethal rampages”. The Florida swimsuit seems to be to carry Altman personally liable, alleging that he has an “utter disregard for the danger to human life”.
One in eight teenagers and younger adults aged between 18-21 turned to AI chatbots equivalent to ChatGPT for psychological well being points, in line with a 2025 research performed by Brown College College of Public Well being, Harvard Medical College and nonprofit analysis organisation RAND.
One other research, from West Texas A & M College additionally focusing on adolescents and younger adults, discovered that almost a fifth of all adolescents developed dependency on AI with beforehand present psychological well being issues as a predisposition for creating the dependency.
Authorized adjustments
Legislators have begun to take word. In Canada a brand new digital safety bill, launched on Wednesday, would require corporations equivalent to OpenAI to be extra “clear” about their reporting requirements when in disaster conditions the place customers could harm themselves or others.
In Washington state, the governor signed a invoice into regulation that requires AI chatbots to remind customers they don’t seem to be human each three hours and is ready to take impact in January 2027. Different states like Illinois, for instance, have banned AI remedy.
On the federal stage, Consultant Mike Lawler, a Republican from New York state, launched a invoice that may require chatbot corporations to inform parents of interactions the place suicidal ideation is mentioned by a person. Nonetheless, this invoice solely applies to minors.
