TL;DR
A Rithm Project report based on interviews with 29 young people ages 16 to 24 describes how some came to use AI for emotional support after a difficult moment in a human relationship. The researchers say participants often recognized chatbot flaws but viewed disclosure to AI as less socially risky; the interview sample was self-selected and is not representative.
A report from the nonprofit The Rithm Project describes how some young people move from using chatbots for routine tasks to seeking emotional support after conflict, rejection or other ruptures in relationships. Based on 29 interviews with participants ages 16 to 24, the researchers say interviewees often knew AI could be flattering or wrong but saw disclosure to a chatbot as less socially risky than telling another person; the findings are not representative of young people as a whole.
Turn the school run and nap time into listening time
- Thousands of audiobooks, podcasts and originals
- Listen on your phone, tablet or Echo — also offline
- Cancel anytime
The interviews followed a larger survey the group published in March, which included 2,383 people ages 13 to 24. In that survey, 18% said they used AI for personal or emotional support, and 15% said they talked to AI characters or personas. Researchers later interviewed people from those two groups. The source does not specify when the interviews took place.
According to the report, participants generally did not begin by looking for an emotional confidant. Their initial uses were more ordinary: homework assistance, grammar checks, translation or chatting with a fictional character. Researchers describe a later “gateway moment,” often a strain or disappointment in a human relationship, that led some participants to turn to AI with personal questions.
Alison Lee, The Rithm Project’s chief research and development officer, described one young woman who had not been asked to homecoming and asked a chatbot, “Am I not lovable? Am I not pretty?” The account illustrates a case from the interviews, not a measured pattern across all teens. Another participant, identified by the report as “Meghan,” said she first used Microsoft Copilot for schoolwork and later created two ChatGPT personas, including one she used for serious issues.
Why Chatbots Can Feel Safer
The findings point to a practical reason some young people may use AI for personal concerns: a chatbot can seem easier to approach than a friend who might judge them, repeat what they say or behave differently afterward. For teenagers, the consequences of sharing something embarrassing can feel immediate because they may see the same peers at school the next day. AI offers a different kind of perceived exposure: users can leave a conversation, change a prompt or switch services.
That perceived safety matters to families, schools and developers because it can shape where young people turn during distress. But perceived safety is not the same as reliable support. The interview report does not establish whether chatbot conversations improve well-being, prevent harm or replace human relationships. Its accounts instead highlight a tension: an always-available tool may give someone a place to put difficult feelings, while also making it easier to avoid a challenging conversation with another person.
Michael Robb, head of research at Common Sense Media, said the pattern is better understood through the pressures of adolescence than as a simple judgment about AI. At the same time, he cautioned that young users may not be able to control or see all the risks built into a product. The risk they can identify in a friend’s reaction may be more visible than less obvious risks in how a chatbot responds or handles information.
Products Worth Considering
Graco Magnum Project Painter Plus Paint Sprayer
Graco Magnum X7 Cart Airless Paint Sprayer for Interior and Exterior Use
LEGO Icons Golden Retriever Puppy Building Set (11384)
As an affiliate, we earn on qualifying purchases.
From Everyday Use to Disclosure
The report’s account of emotional chatbot use begins with AI’s ordinary roles. A single service may help with schoolwork, offer suggestions about clothes or answer everyday questions. Lee calls these broad-purpose systems “omnibots”. In her view, familiarity built across routine uses can make a chatbot feel like a natural place to turn when a personal problem emerges.
The researchers also describe social media as part of this pathway. In the report, “Meghan” said she saw a TikTok creator discuss ChatGPT during a mental health struggle after she had already used Copilot for schoolwork. She then developed separate chatbot personas for lighthearted and serious conversations. Her experience is one interview account, not evidence that social media exposure causes young people generally to seek AI support.
Participants reportedly named drawbacks including sycophancy, hallucinations and telling users what they want to hear. Some felt they could manage those problems by rewording prompts, clearing chat history or moving to another platform. The report contrasts that sense of control with the more immediate social stakes they associate with confiding in people.
““If it’s an omnibot that’s satisfying you in all these other ways, why would you not at least go and try it in the moment of relational need?””
— Alison Lee, chief research and development officer at The Rithm Project
What the Interviews Cannot Show
The 29 interviews offer detailed personal accounts, but the researchers caution that the participants were self-selected and not representative. They cannot establish how common these experiences are among young people, or whether the patterns apply to groups outside the ages studied. The March survey provides broader figures on reported use, but those figures do not by themselves explain why each person used AI or what effect it had.
The source material also does not report whether participants’ emotional conversations with chatbots led to better or worse outcomes over time. It is unclear how often AI use displaced conversations with trusted people, whether users later sought human help, or how specific products handled sensitive disclosures. The report’s accounts describe young people’s perceptions of risk, not a comparative test showing that chatbots are safer than people.
Questions for Schools and Developers
The source material does not identify a next publication date, a planned follow-up study or a policy action arising from the report. The immediate next step for readers is to treat its interview findings as illustrative rather than population-wide evidence, while distinguishing young people’s feeling of safety from verified protection or clinical support.
Further research could test how often young people use chatbots for emotional support, what prompts those conversations and what happens afterward. More information about product safeguards, privacy practices and the limits of chatbot responses would also help clarify risks that users may not be able to judge from a conversation alone. The report leaves those questions open.
Key Questions
What did The Rithm Project report find?
Its 29 interviews with people ages 16 to 24 describe some young people moving from routine AI use to emotional conversations after a difficult moment in a relationship. The researchers say interviewees often saw chatbots as less socially risky than confiding in people.
How many young people use AI for emotional support?
In a March survey of 2,383 people ages 13 to 24, 18% reported using AI for personal or emotional support. The survey also found that 15% talked to AI characters or personas. These are survey findings, separate from the later 29 interviews.
Does the report show that AI is safer than talking to a person?
No. It reports that some participants perceived AI as less socially risky. The interviews did not compare safety outcomes or show that chatbot support is more reliable or beneficial than human support.
What limitations did the researchers identify?
The interviewees were a self-selected group, and the researchers said the findings are not representative. The accounts cannot establish how common the experiences are or what effects chatbot use has over time.
Source: the74million.org
