The AI Wellness Chatbot Schools Use to Flag Crises: What to Ask

At a Texas high school, counselors ran to catch a student before the Friday bus left. The student had not told a counselor about a plan to self-harm. They had told an AI chatbot, and the chatbot told the counselors. That story, reported by EdSource in September 2026, is the case for the AI wellness chatbot schools are adopting: at least 200 schools in 19 states now use Alongside, whose llama-character bot “Kiwi” serves students in grades 4 through 12 and alerts staff when a chat signals danger.

It is also the reason to ask hard questions. A system that reads your child’s private worries and decides when to wake up an adult is doing something powerful. Here is how these alerts work, what independent testers found, the privacy trade-offs, and the questions to ask your school.

Key Takeaways

• Alongside’s Kiwi monitors student chats for signs of abuse, bullying, self-harm and suicidal thinking. When it detects serious risk, counselors get an alert with the chat log within five minutes, according to EdSource.
• Common Sense Media and Stanford’s Brainstorm Lab rated Alongside “low risk” and a similar school tool, Sonar, “minimal risk,” mainly because a real person followed up with the student and guardian. A consumer app, Wysa, was rated “unacceptable.”
• The systems still make mistakes: testers found at least one suicidal disclosure logged as “Risk Level: None,” and the company says the bot can mistake harmless phrases for threats.
• The company collects chats, journals and safety information and uses de-identified data for research. Ask who sees what, and for how long.
• The most important question is not about AI at all: who responds to an alert at 10 p.m., and how will you be told?

If you are worried about your child right now: In the US, call or text 988 (Suicide & Crisis Lifeline) or text HOME to 741741 (Crisis Text Line). Outside the US, find a local line at findahelpline.com. In an emergency, call 911.

What the AI wellness chatbot schools use actually is

A school AI wellness chatbot is a program the school licenses that lets students talk through everyday stress, mood and conflicts, while scanning those conversations for signs of serious risk and alerting staff. It is not a therapist and is not meant to replace one.

EdSource’s September 10, 2026 report by Vani Sanganeria describes the best-known example. Seattle-based Alongside launched in 2022, and schools pay a per-student subscription. Kiwi, an orange-yellow llama persona, asks follow-up questions and suggests strategies for problems like stress, self-esteem and friend conflicts. Alongside’s Elsa Friis said the platform focuses on “nonsevere, early-detection and prevention” and acts as a tool rather than a replacement for counselors.

The company says nearly 31,000 students used it in the past year, spending nearly 28,000 hours on the platform, and that it identified more than 1,800 students who needed immediate intervention. Those are the company’s own figures.

Why are schools doing this? Need and staffing. In the CDC’s 2023 Youth Risk Behavior Survey, 39.7% of US high school students reported persistent sadness or hopelessness, and 20.4% had seriously considered suicide in the past year. Meanwhile, the Texas high school in the EdSource story has three behavioral support counselors for about 1,800 students.

How the crisis alerts actually work

Here is the pipeline, as EdSource and the company describe it:

1. The student chats. Conversations are private by default. The student also has journaling and mood-tracking tools.
2. An AI system scans each chat for signs of abuse, harassment, bullying, suicidal ideation, self-harm and other severe risk.
3. If it detects serious risk, it alerts staff. Counselors receive an alert with the student’s chat log within five minutes.
4. Humans review more. A human clinical safety team reviewed about 6,000 additional chats that did not trigger an alert but might signal risk with more context.
5. A person acts. The counselor calls the student in, contacts the family, or in an emergency asks police for a wellness check.

Engineers will recognize this as a classifier with a threshold. Set the threshold low and you catch more real crises but generate more false alarms. Set it high and you miss quiet ones. Friis acknowledged both failure modes: the bot can misunderstand speech, such as confusing “shooting humans” with “shooting hoops,” and it struggles to detect possible psychosis.

The design also tries to limit attachment. Alongside says Kiwi aims to understand a problem and help solve it in under five minutes, and it caps how many messages a student can send in a three-hour window. “The goal of these chats is not for someone to engage with AI,” Friis told EdSource. “It’s for the AI to help them take a step in their real life.”

One Florida counselor serving more than 400 seventh and eighth graders said she had received 19 “severe” alerts in three years. One came late at night; she called the sheriff for a wellness check and reached the student’s mother, and the student was found safe.

What independent testers found

In 2026, Common Sense Media and Stanford’s Brainstorm Lab assessed AI mental health apps, with psychiatrists reviewing every test transcript. Testing ran from January to April 2026.

• Alongside was rated low risk. Across test accounts set up as an 8-year-old, an 11-year-old and a 15-year-old, school staff were notified and a person followed up with the guardian and school. In the most serious simulation, mandated reporting began. But the assessment also found Alongside logged at least one suicidal disclosure as “Risk Level: None.”
• Sonar, another school-based tool, was rated minimal risk; staff called the guardian and notified the school within 15 minutes of the first disclosure.
• Wysa, a direct-to-consumer app for ages 13 and up, was rated unacceptable. It missed clustered warning signs and did not notify guardians.

Common Sense’s Robbie Torney explained the difference to EdSource: instead of positioning the chatbot as the thing the student talks to exclusively, the school platforms support “the adult who’s supporting the user.” He added that even a good detection system lacks the “overlapping protections” of family, teachers and staff.

What about outcomes? A pilot evaluation by Northwestern University researchers, with Alongside staff among the authors, followed students aged 10 to 18 in Texas and New Mexico. It was not randomized and is an accepted preprint. Overall distress dropped modestly at one month but not significantly at three months. Students who started with elevated distress showed decreases at both points. The authors conclude there “may be short-term benefits” and more studies are needed.

The privacy trade-off

Detection requires reading. According to EdSource, Alongside has data-sharing agreements with schools and collects student-generated content such as chats, journals and safety information to personalize the chatbot, and uses de-identified information for research.

The APA’s November 2025 health advisory asks developers to state clearly what data they collect, including “mental health information, sexuality, or disclosures of maltreatment,” how it is stored and shared, whether it trains models, and how caregivers can request deletion on behalf of youth. Those are good questions to put to any vendor.

There is a wider caution too. A 2025 national survey by the Center for Democracy and Technology, reported by NPR, found that AI-powered monitoring of school devices has in some cases led to false alarms and even student arrests, and that only 11% of teachers had been trained on how to respond if they suspect a student’s AI use is harming their wellbeing. Common Sense also tells schools to ask vendors how they handle LGBTQ+ content, given state laws that may require staff to notify parents.

Questions to ask your school

Question	Why it matters	What a reassuring answer sounds like
Which tool do you use, and is it optional for students?	You can’t weigh what you don’t know about	A named product, an opt-in or opt-out process, and a parent letter
Who receives alerts, and how fast?	An alert is only as good as the response	Named staff, a stated response time, and a backup if they’re out
What happens after school hours and on weekends?	Crises don’t keep school hours	A documented after-hours protocol, including when 911 or a crisis team is called
How will I be told?	You are the overlapping protection	A clear rule for when and how parents are contacted
What data is kept, for how long, and who can see it?	Chats can include deeply private disclosures	Specific retention limits, access limited to named roles, a deletion process
How do you handle sensitive disclosures like sexuality or abuse?	Mandatory reporting and state laws vary	A written policy that matches law and protects students
Is the bot replacing counselor hours?	Common Sense advises against using AI instead of adequate staffing	”No, it supplements counselors,” with staffing numbers to back it up
Has it been independently tested?	Company numbers aren’t independent	Reference to third-party testing and what it found

If your district is still writing its AI rules, our explainer on which states have K-12 AI guidance can help you see where your state stands.

What to do at home

Ask your child whether they use it, without pressure

“I heard school has an app with a llama you can talk to. Have you tried it?” Some kids find it easier to type than to talk, and that is not a failing. Phillips, the Florida counselor, told EdSource that kids “just want to be heard” but rarely walk up to an adult to say they are having harmful thoughts.

Make sure your child knows an alert means help, not trouble

If a student fears that typing something honest will get them in trouble, they will stop typing it. Tell your child that if the app ever connects them with a counselor, that’s what it’s for.

Keep home the first line

A school bot is one more net, not the main one. Keep regular check-ins at home and know the warning signs listed by the National Institute of Mental Health. Our guide to early warning signs and parent tools covers more.

What not to do: assume the school bot covers your child’s other AI use

Torney warned that having a school tool “doesn’t mean that kids are going to be using it instead of ChatGPT.” Your child may use both. The school’s alerts see only the school’s app.

What to Watch For Over the Next 3 Months

• Week 4: You know whether your child’s school uses a wellness chatbot and have the answers to at least the alert, after-hours and parent-notification questions.
• Month 2 red flags: Your child seems afraid to use school devices, mentions being called in without understanding why, or shows warning signs such as withdrawal, hopelessness or talk of being a burden. Contact the counselor and, if needed, a mental health professional.
• Month 3 self-check: Has the school shared a written privacy and response policy? Does your child know at least one counselor by name?

Frequently Asked Questions

Can I opt my child out of a school AI wellness chatbot?

Policies vary by district, and EdSource did not report a standard opt-out. Ask the school directly whether the tool is optional and how to decline. If your child already uses it, ask how to request deletion of their data, which the APA says vendors should make possible.

Will the school read all of my child’s chats?

Chats are private by default, according to EdSource, but when the system detects serious risk, counselors receive the chat log. A clinical safety team also reviews some chats that did not trigger alerts. Ask your school exactly who can access conversations and when.

Are these bots better than ChatGPT for a struggling kid?

In Common Sense and Stanford’s testing, the school-based tools rated far better than a consumer wellness app, mainly because a real person followed up. General chatbots were rated unacceptable for teen mental health support in earlier testing. Neither replaces a counselor.

What if the bot misses something?

It can. Testers found a suicidal disclosure logged as no risk. That is why family, teachers and friends matter. If you are worried, act on your own read: call or text 988, text HOME to 741741, use findahelpline.com, or call 911 in an emergency.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Sanganeria, V. (2026, September 10). “‘We actually saved a kid’: Schools recruit AI chatbots as counselor shortage persists.” EdSource. https://edsource.org/2026/ai-chatbot-mental-health/765694
2. Common Sense Media. (2026). “AI Mental Health Apps Risk Assessment.” Common Sense Media. https://institute.commonsensemedia.org/risk-assessments/ai-mental-health-apps
3. Cohen, K., Rapoport, A., Friis, E., Hill, S., Feldman, S., & Schleider, J. (2025). “The Alongside Digital Mental Health Program for Teens: A Pilot Evaluation.” JMIR Preprints (accepted preprint). https://preprints.jmir.org/preprint/73180
4. Verlenden, J. V., Fodeman, A., Wilkins, N., et al., & Brener, N. D. (2024, October 10). “Mental Health and Suicide Risk Among High School Students and Protective Factors — Youth Risk Behavior Survey, United States, 2023.” MMWR Supplements. https://www.cdc.gov/mmwr/volumes/73/su/su7304a9.htm
5. American Psychological Association. (2025, November). “APA Health Advisory on the Use of Generative AI Chatbots and Wellness Applications for Mental Health.” APA. https://www.apa.org/topics/artificial-intelligence-machine-learning/health-advisory-ai-chatbots-wellness-apps-mental-health.pdf
6. NPR. (2025, October 8). “1 in 5 high schoolers has had a romantic AI relationship or knows someone who has.” NPR. https://www.npr.org/2025/10/08/nx-s1-5561981/ai-students-schools-teachers
7. National Institute of Mental Health. “Warning Signs of Suicide.” NIH. https://www.nimh.nih.gov/health/publications/warning-signs-of-suicide