Newsroom
Weekly ReviewNovember 5, 2026

When Students Turn to AI in Distress: What Schools Can Do

A JED survey of 5,560 students found 57% used AI when sad or stressed. What the data show on chatbots and student mental health, and what schools can do.

In a survey of 5,560 middle and high school students across 21 U.S. schools, 57% said they had used an AI tool when they felt sad, stressed, lonely, or overwhelmed. The Jed Foundation (JED) released those findings this fall, days after a nationally representative Common Sense Media census found that 65% of 9- to 17-year-olds use AI in their free time. For school leaders, the open question about AI chatbots and student mental health is no longer whether students use these tools, but what a school does when a chatbot becomes the first place a struggling student goes.

What the JED student survey measured

JED's report, Exploring Artificial Intelligence: Understanding Relationships between AI Use and Mental Health Among Middle and High School Students, draws on its Student Survey, fielded from October 2025 through April 2026. Responses were weighted for non-response and analyzed with models adjusted for sex, race/ethnicity, and grade.

Two limits shape how the findings should be read. The sample is not nationally representative; it comes from 21 participating schools. And the data are cross-sectional, collected at one point in time, so the report cannot say whether AI use contributes to poorer well-being or whether students who are already struggling are more likely to turn to AI. JED states both limits directly.

Within those limits, the usage picture is clear:

  • 81% of students reported using AI in some way.
  • 58% used it for schoolwork or homework.
  • 28% used it for ideas or advice about relationships or problems.
  • 12% used it for friendship or companionship.
  • 57% used it at a time when they felt sad, stressed, lonely, or overwhelmed.
  • 8% said they directly asked AI for help with those feelings.

Students could select more than one reason, so the categories overlap.

How AI chatbot use relates to student mental health in the data

The report's central finding is that how students use AI matters more than whether they use it. According to JED, using AI in general was not associated with meaningfully different well-being scores compared with not using it.

Specific uses told a different story. Students who used AI for companionship, and students who asked AI for help with their feelings, reported higher anxiety and depression symptom scores, more loneliness, and lower resilience and flourishing scores than non-users. Using AI for advice was associated with higher anxiety and depression symptom scores and more loneliness. Using AI for schoolwork showed a small association with loneliness and no significant association on the other measures.

The associations were especially pronounced among students who said they turned to AI because "I don't have anyone to talk to" or "to feel less alone." That detail matters for schools. It suggests that for some students, a chatbot is filling a gap left by the absence of a trusted person, not competing with one.

These are associations. They do not show that chatbots cause distress, and they do not show that removing access would improve well-being. They do identify a group of students whose AI use is a signal worth paying attention to.

Middle school students and girls reported more distress-related use

JED found clear differences by grade and sex:

  • Grade: 65% of students in grades 6–8 reported using AI when emotionally struggling, compared with 49% in grades 9–10 and 50% in grades 11–12.
  • Sex: 62% of female students reported distress-related use, compared with 51% of male students. Female students were also more likely to report companionship use (15% vs. 9%) and directly seeking help with feelings (11% vs. 5%).
  • Race/ethnicity: Distress-related use was reported by 72% of American Indian/Alaska Native students, 65% of Black and of Hispanic students, 64% of Asian students, 56% of multiracial students, and 47% of White students.

JED flags these patterns as having implications for mental health disparities. For district planning, the grade finding stands out: students in grades 6–8 reported distress-related AI use more often than students in any high school grade band.

What the national data and professional guidance add

Common Sense Media's 2026 Common Sense Media Census: Tech and Media Use by Tweens and Teens, released October 27, 2026, describes a nationally representative survey of 9- to 17-year-olds. It found that 65% use AI in their free time (77% of teens and 49% of tweens), and that 49% of tweens have used an AI chatbot or companion. The census measured use outside school and homework by design. That is a useful reminder that a district's acceptable-use policy governs only part of a student's AI exposure.

The American Psychological Association has issued two advisories relevant here. Its June 2025 health advisory on AI and adolescent well-being calls for comprehensive AI literacy education, encourages adults to help adolescents sustain supportive relationships with other people, and recommends that students be directed to a trusted person, such as a school counselor, to verify health information they encounter through AI. Its November 2025 advisory on generative AI chatbots and wellness apps advises against using these tools as a substitute for care from qualified mental healthcare professionals and calls for specific safeguards for children and teens.

What this means for district and school leaders

JED's report includes recommendations for youth-serving professionals, including schools. Translated into district practice:

Write AI guidance by function and risk. JED recommends institutional guidelines matched to what a tool does and the risk it carries. An approved writing assistant and an open-ended chat feature that responds to emotional disclosures are different tools. Review district-approved platforms for conversational or companion features, and decide explicitly how those features are handled.

Make the route to a person visible. JED recommends making clear how students can reach human support, including on-site staff and referral pathways. In practice, that means students should be able to name who to go to, and how, without having to ask. Post it where students already look, including the digital spaces they use for school.

Check that the route leads somewhere. A visible pathway only helps if someone can respond. Earlier weekly reviews covered CDC data on unmet counseling need among high school students and RAND findings on the time teachers already spend on student well-being. Any plan to redirect students from chatbots to people should be paired with an honest look at staff capacity.

Teach AI literacy with real situations. Both JED and APA recommend AI literacy that goes beyond technical skills. JED specifically names emotional distress, advice-seeking, and privacy as scenarios to cover. This fits naturally into Tier 1 instruction within an existing MTSS framework, alongside SEL and digital citizenship work.

Invest in belonging. JED's first recommendation for youth-serving professionals is to strengthen connection with peers and trusted adults through mentoring, arts, sports, and community activities. Given the "no one to talk to" finding, this is not a side project to AI policy; it is the core of it.

Bring students and families into tool decisions. JED recommends involving young people and caregivers in selecting and evaluating AI tools. A short family-facing explanation of which tools the district uses, and why, also gives parents a starting point for the conversations JED encourages at home.

Plan for middle school. If capacity is limited, the grade-level pattern in JED's data suggests middle school is a reasonable place to start AI literacy and connection work.

Looking ahead

AI tools will keep changing faster than district policy cycles, and the research on adolescents is still early and largely cross-sectional. What the current evidence supports is narrower and more durable: students who turn to chatbots because they have no one else to talk to are the students schools need to reach first, and the response is a person, not a product setting. School-Based Healthcare Solutions Network, Inc. (SBHSN), a private grant-making organization focused on access to mental health care on school campuses, publishes this weekly review to help school leaders turn national and state developments into local decisions. Schools and districts weighing campus-based mental health capacity can review the grant solicitation and its published criteria at sbhsnetwork.com/grants. Application deadlines fall on the first Monday of July, September, November, and February; questions can be sent to grants@sbhsnetwork.com.

Sources

Transitional Coach℠ and Coaching Model℠ are service marks of School-Based Healthcare Solutions Network, Inc. © 2026 School-Based Healthcare Solutions Network, Inc. All rights reserved. This article is provided for general informational purposes only and does not constitute clinical, legal, or financial advice.