Do teens who use AI for companionship really face the highest rates of suicidal thoughts?

01The short version

• JED surveyed 5,560 middle and high school students in 21 U.S. schools. Those who use AI as a friend or companion (12% of students) were about twice as likely to report a lifetime history of suicidal thoughts.
• It is a one-time survey, so it cannot say whether AI made things worse or struggling teens turned to AI. JED says so itself.
• The word "highest" is shaky. Students who use AI to get help when sad or stressed show a bigger gap, and using AI for any reason overall showed no significant link.

If you or someone you know is struggling, you can call or text 988, the Suicide & Crisis Lifeline (US), any time.

02Where the claim came from

On October 27, 2026, TechTimes ran a story saying teens who use AI for companionship face the highest rates of suicidal thoughts.³ It was based on a new report from The Jed Foundation (JED), a youth suicide-prevention nonprofit. NPR covered the same report three days later.²

JED’s report is a survey of 5,560 middle and high school students in 21 U.S. schools, collected between October 2025 and April 2026. Students were asked why they use AI, and how they feel. The data were adjusted for sex, race and grade. The sample is not nationally representative.¹

03What the evidence actually shows

05What the headline gets right

The study exists and says roughly what the headline says. Students who used AI for friendship or companionship had more anxiety and depression, more loneliness, lower resilience, and a greater history of suicidal thoughts and attempts than students who did not use AI that way.¹ A gap of about two to one is large. JED’s authors call these among the most consistent findings in the report, and they ask companies to keep AI companions away from minors.¹

It is also not a one-off. An early, unpublished study from the University of North Carolina found middle schoolers who used chatbots for companionship grew lonelier over time.⁴

06What it gets wrong or leaves out

“Highest.” Companionship is not the top group in JED’s tables. Students who turn to AI to get help when they feel sad, stressed or lonely show a bigger gap for suicidal thoughts (2.43 against 2.01) and for attempts. Some reasons within that group, such as having no one else to talk to, are higher still.¹ JED itself groups the two together as its most consistent findings.¹ The TechTimes story says companionship was worse than every other category “by a significant margin,” which JED’s tables do not show.³

“Face.” The headline reads like a forecast: use AI this way and this is what you will face. The survey is a single snapshot. Kayla Anderson, JED’s vice president of research and evaluation, says it is unclear whether AI use contributes to poor mental health or whether students already in distress are more likely to turn to AI.² The report says the same, and says the sample is not nationally representative.¹ To its credit, the TechTimes story repeats that caveat further down the page.

“Rates of suicidal thoughts.” The measure is a lifetime history of suicidal thoughts, not current thoughts. A teen who had such thoughts years ago counts.¹

What is missing. Using AI for any reason was not linked to worse mental health overall.¹ The finding is about how teens use AI, not whether they do.

07Where the real concern is

Many teens already turn to AI when they are hurting. In JED’s survey 57% of students used AI when they felt sad, stressed, lonely or overwhelmed, though only 8% said they were using it to “get help.”¹ A separate nationally representative survey found about one in eight young people use chatbots for mental health advice.⁵

Whichever way the link runs, some teens who are struggling are reaching for a product that was not built or tested for that job. That is a design and oversight question for companies and lawmakers, and it does not depend on proving the chatbot caused the harm.

08What we’re not sure about

Direction. We do not know whether companionship use worsens mental health, whether struggling teens seek companions, or both. Tracking the same teens over time would answer it, and the UNC study that does so is not yet published.⁴ We also have only JED’s report and its own tables: it has not been through outside peer review, and the survey could not account for how a student was doing before they started using AI.