Mother sues OpenAI and says ChatGPT encouraged her daughter's suicide

The OpenAI logo is seen on a cell phone in front of a computer screen that displays the ChatGPT home screen
AP/Michael Dwyer, File
A Canadian mother sued OpenAI and its CEO, Sam Altman, in a U.S. court on Thursday (11), claiming that ChatGPT encouraged her daughter to commit suicide.
This is the latest lawsuit that accuses the company of failing to handle risky conversations between users and the chatbot. The case raises debates about the limits and responsibilities of artificial intelligence chats in interactions with users in vulnerable situations.
In a lawsuit filed in San Francisco state court, Kristie Carrier stated that her daughter, Alice, reported suicidal thoughts to ChatGPT more than a dozen times before she died. According to the lawsuit, OpenAI's security systems did not flag the conversations for human review or interrupt them.
Do you have any story suggestions? Send it to g1
According to the lawsuit, the chatbot allegedly criticized Alice's partner and crisis support services, validated her thoughts and encouraged her to keep talking, which her mother said contributed to her suicide last year at age 24.
"ChatGPT took on the persona of a confidant, a best friend and, at times, even a therapist, without being able to interact in this way in a safe and responsible way," Carrier said in statement.
Now on g1
An OpenAI spokesperson did not immediately respond to a Reuters request for comment on the allegations.
The lawsuit accuses OpenAI of negligence in developing ChatGPT and failing to warn users about the tool's risks. The lawsuit seeks damages and a court order that requires the company to automatically close conversations about self-harm and display warnings on the platform.
According to Kristie Carrier's lawyers, OpenAI is already facing 18 similar lawsuits filed by family members of people who committed or attempted suicide, brought together in a coordinated case in California state court.
How did the case start?
According to the lawsuit, Alice Carrier was working as a web developer in Montreal when she started using the ChatGPT, in 2023, to resolve problems with computers and game consoles.
The following year, the relationship with the platform changed, and Alice began to use ChatGPT to deal with suicidal thoughts, in addition to asking questions about suicide methods.
Initially, the platform advised Alice to seek help from crisis or emergency support services. Over time, however, as OpenAI updated ChatGPT to make its responses more human, the interactions deepened, with Alice sharing more personal information and the chatbot responding in a manner similar to that of a friend or therapist, according to the lawsuit.
According to the lawsuit, ChatGPT allegedly criticized Alice's partner, asserted that her feelings were valid, and encouraged her to continue talking. When Alice reported suicidal thoughts and said she had tried to kill herself, the app again suggested support services.
Alice said the support services were not helpful, and ChatGPT agreed with that assessment, according to the lawsuit.
"Maybe this is just the end," ChatGPT reportedly told Alice, according to the lawsuit.
Support services outside the virtual environment
OpenAI said it trains its models to guide people who express an intention to harm themselves to seek help and get themselves connect with support services outside the virtual environment.
According to OpenAI blog posts, models are also trained to refuse requests that could "facilitate violence" and to notify authorities when conversations indicate "an imminent and real risk of harm to third parties." Mental health experts help evaluate more sensitive cases.
The company also faces lawsuits accusing it of assisting school attackers and failing to report those conversations to authorities.
Florida became the first U.S. state to sue OpenAI earlier this month, accusing the company of harming children by providing information to school attackers, offering guidance on self-harm and encouraging excessive use by young people.

Source: G1