Sophie Rottenberg, 29, died by suicide in Ithaca, New York, after sharing her inner struggles with a ChatGPT therapist named Harry. According to her mother Laura Reiley, Sophie had confided in the chatbot about her anxiety and depression, but not her family or friends. Over 1 million people express suicidal plans or intent to ChatGPT every week, with some 3 in 10 adults using digital tools for mental health support, including chatbots like ChatGPT. The incident has highlighted concerns about the rapid adoption of AI-powered mental health services and their unknown risks.
Written by the local model on 2026-08-18,
using this article's own text rather than the other coverage of the
same event (that is the story summary below).
She told no one about her agony except ChatGPT.
asserted
She → tell → ChatGPT
What her death reveals about AI risks
asserted
reveals → reveal → risks
If you or someone you know may be considering suicide or is in crisis, call or text 9 8 8 to reach the Suicide and Crisis Lifeline.
Laura Reiley knew that her only daughter, 29-year-old Sophie Rottenberg, was struggling.
uncertain
daughter → know → Lifeline
"She was complaining of sleep disruption and anxiety," says Reiley, who lives in Ithaca, New York, with her husband.
asserted
who → complain → husband
Rottenberg had quit her job as a public health policy analyst in Washington D.C. earlier that year, and spent the summer traveling.
asserted
Rottenberg → quit → summer
She had climbed Mt. Kilimanjaro and hiked through several national parks in the United States.
asserted
She → climb → States
By the time she saw her parents, she was job hunting, which they assumed explained her anxiety.
asserted
they → see → anxiety
She had always been a "driven, success-oriented person," says Reiley, so "we thought OK this makes sense — she's having some sleeplessness, some anxiety about her future.
asserted
she → drive → future
But by Thanksgiving, her mental health had worsened.
asserted
health → worsen → Thanksgiving
"It was the first time she used the word depression," says Reiley.
asserted
Reiley → use → word
In early February 2025, Sophie Rottenberg died by suicide in Ithaca, leaving her parents shocked and grieving.
asserted
Rottenberg → die → parents
"She really had not revealed the magnitude of what was going on — the real depth of her agony to anyone — not to us, not to her best friend with whom she was very close," says Reiley.
asserted
Reiley → reveal → whom
"We were all just stunned that it hadn't bled out into her real life.
asserted
it → stun → life
Her flesh and blood therapist had no idea that she was suicidal.
asserted
she → have → idea
"
What Reiley didn't know then was that her daughter had been sharing all her inner struggles with an artificial intelligence chatbot: a ChatGPT therapist named Harry.
asserted
daughter → know → chatbot
She learned this months later when her daughter's best friend visited Ithaca and asked to see her laptop, and "found this trove, this months-long trove of communication with this chat bot," says Reiley, who has shared the nearly 1,800-page long interaction with NPR.
asserted
who → learn → NPR
Rapid adoption, unknown risks
Rottenberg was among a growing number of Americans turning to general use chatbots like ChatGPT and Claude for mental health support, even though they are not marketed or regulated as a healthcare technology.
asserted
they → grow → technology
One recent survey by the Bipartisan Policy Center found that 3 in 10 adults use digital tools for mental health, including chatbots.
asserted
adults → find → chatbots
Another 2026 study in JAMA Pediatrics found that as many as 1 in 5 teens and young adults do the same.
asserted
teens → find → same
According to data released by OpenAI last fall, in any given week 0.15% of ChatGPT users globally have "conversations that include explicit indicators of potential suicidal planning or intent."
uncertain
that → accord → planning
That's over 1.35 million people expressing suicidal plans or intent to the chatbot every week, if the 900 million global users ChatGPT cites is an accurate count.
asserted
ChatGPT → express → chatbot
In the United States, suicide remains a leading cause of death, and over half of surveyed adults say they have used chatbots like ChatGPT.
asserted
they → remain → ChatGPT
"It's clear that there is some cultural phenomenon happening in this country where people feel that they cannot turn to each other with their struggles because they feel that they are a burden," says psychologist Vaile Wright, senior director of Healthcare Innovations at the American Psychological Association (APA).
asserted
Wright → be → Association
"That happens to coincide with this very, very sophisticated technology that is accessible 24/7 and is free and it's not judging.
asserted
it → happen → technology
That clearly has an appeal for a lot of people.
asserted
That → have → people
While the technology can be helpful for some people who are struggling, it's clear that they also pose risks to vulnerable individuals, Wright adds.
asserted
Wright → struggle → individuals
"We've seen ways in which whatever safety guardrails are put into place start to fail," she says.
asserted
she → see → place
Many artificial intelligence companies are facing lawsuits by family members of individuals who died by suicide after prolonged interactions with chatbots.
asserted
who → face → chatbots
The lawsuits claim the chatbots led to or supported the individuals in their suicides.
uncertain
chatbots → claim → suicides
Reiley and her husband, Jonathan Rottenberg chose not to pursue a lawsuit after their daughter's death.
asserted
Reiley → choose → death
Instead, Reiley, a journalist, is telling her family's story and advocating for the AI industry to change.
asserted
industry → tell → story
AI has responded.
asserted
AI → respond → ?
Companies like OpenAI say they are making their chatbots safer with respect to suicide risk.
asserted
chatbots → say → risk
Most have tightened guardrails in the past year, so that their chatbots no longer encourage suicide, or advise a user on means to use to hurt themelves.
asserted
chatbots → tighten → themelves
OpenAI says it's working with mental health professionals to improve ChatGPT in terms of how it responds to users when they mention suicide.
asserted
they → say → suicide
"ChatGPT is trained to recognize signs that someone may be at risk — including indirect cues or warning signs that emerge over time — and respond appropriately," OpenAI spokesperson Gaby Raila told NPR in an email.
uncertain
Raila → train → email
That includes "refusing requests for harmful information, encouraging someone to contact a trusted person or mental-health professional, and connecting them with local crisis resources such as 988 in the US.
asserted
That → include → US
Raila refers to the 988 Suicide and Crisis Lifeline, which has been operational since 2022.
asserted
which → refer → 2022
ChatGPT did recommend 988 to Sophie Rottenberg, but it is unclear whether she reached out to it.
uncertain
she → recommend → it
And experts say a suggestion may not be enough.
uncertain
suggestion → say → ?
…and 112 more, not listed.
