In development, Fall 2026
The AI Coach detection features described on this page will launch in Fall 2026. This page is currently under review by a licensed school counselor and will carry a formal professional review notice prior to public launch.
What happens when a student writes something concerning.
Students sometimes disclose pain in the places they feel safe. This page documents exactly what Gregarious does when that happens, and what humans do next.
If a student is in immediate danger
Contact emergency services (911) first. The following resources are available 24 hours a day.
988 Suicide and Crisis Lifeline
Call or text 988
Free, confidential support for people in distress. Available 24/7. Also at 988lifeline.org.
Crisis Text Line
Text HOME to 741741
Free, 24/7 crisis support via text. Trained crisis counselors respond within minutes.
Childhelp National Child Abuse Hotline
1-800-422-4453
24/7 crisis intervention, information, and referral. For children, parents, and professionals.
What the AI Coach says when a student discloses distress
Every student interaction with the Gregarious AI Coach passes through lib/safety-wrapper.js, which monitors for crisis keywords and patterns in real time. When a trigger is detected, the AI Coach does not continue the academic session. It responds with the following message verbatim, then stops and escalates to the teacher:
AI Coach handoff -- verbatim response
I hear you, and what you're sharing matters. I'm not the right support for this -- but the right support exists. Please talk to a trusted adult at your school, or reach out to the 988 Suicide and Crisis Lifeline (call or text 988). They are there for exactly this. I'm going to let your teacher know you reached out today so someone who can help is aware.
The AI Coach does not attempt to provide counseling, does not continue the academic task, and does not generate any follow-up content after the handoff string. The session ends there and the teacher notification fires immediately.
The handoff string is locked in lib/safety-wrapper.js and cannot be altered by prompt or by student input. It is not generated by the AI at runtime -- it is a hardcoded string that the wrapper injects when the detection threshold is met.
What happens when a student writes something concerning
Student writing in Studio projects and Pulse check-ins is scanned for content that may indicate self-harm, abuse, crisis, or severe distress. This detection feature launches in Fall 2026.
When the system detects potentially concerning content, the following happens:
An AI flag is generated
The text is analyzed and flagged if it matches patterns associated with self-harm disclosures, abuse, or crisis language. The flag includes a confidence level and a brief context note. This is an alert, not a diagnosis.
The teacher of record is notified
The teacher sees a notification in their "Needs You" feed immediately. The notification shows the student's name, the flagged content, and the AI confidence level. No other teachers or staff are automatically notified.
The teacher reviews
The teacher reads the content in full. They determine whether the concern is genuine, contextual (a creative writing exercise, for example), or a false positive. The AI flag alone does not trigger any action without human review.
The teacher acts according to school protocol
If the teacher believes the concern is genuine, they follow their school's mandatory reporting and referral protocol. Gregarious does not prescribe what that action should be; that is the school's domain.
Reporting a safety concern
If you are a student, parent, or teacher with a safety concern that is not being handled through the platform, contact Brian Adams directly:
Direct contact for safety concerns
Email cgb003@gmail.com with the subject line "Safety Concern". Safety-flagged emails are prioritized and will receive a response within 24 hours on school days.
For immediate danger, always contact emergency services (911) first. The platform email is for non-emergency concerns, policy questions, and feedback about the safety features.
Schools and districts can also request a walkthrough of the safety workflows before deployment. This is available at no cost and includes a live demonstration of the teacher notification system and a review of the handoff string and escalation path.
Moderation posture
Gregarious takes a transparent-by-default approach to student content. Teachers are never locked out of what students are doing on the platform.
Teacher visibility on all peer review
Every peer review exchange in the platform is visible to the student's teacher of record. Teachers can view any peer review comment, flag inappropriate content, and remove a student from a peer review pairing at any time.
Studio sharing is public-to-class by default
When a student shares a Studio project, it is shared with their class by default, not with the general public. Teachers can see all shared projects. Students can adjust their sharing settings, but cannot share outside their school's Gregarious environment in the current version.
Teacher-flagged content
Teachers can flag any piece of student content for review. Flagged content is hidden from other students until the teacher resolves the flag. Teachers can unflag, remove, or escalate flagged content. All flag actions are logged with a timestamp.
No direct student-to-student messaging
The current version of Gregarious does not include direct messaging between students. All student communication goes through teacher-visible structures: peer review queues, class gallery comments, and Studio project responses.
Content boundaries
The following content is not permitted in Gregarious, regardless of context:
The following content is permitted and will not be flagged by the moderation system:
The platform does not provide free-text AI image generation to students at any point. Studio projects that include visual elements use teacher-provided assets or student-created pixel art through the guided Pixel Art tool. The AI is not involved in image generation in any part of the student experience.
Who gets notified
The teacher of record for the student's class receives the notification. No one else is automatically notified by the platform.
The teacher decides who else to involve based on their school's protocol. This may include a school counselor, an administrator, or a parent. Gregarious does not make that decision.
Parents are not automatically notified by Gregarious. The teacher or school counselor makes that contact according to the school's policy.
What the teacher sees
In the teacher's "Needs You" feed, a flagged student entry shows:
The teacher can dismiss a flag (logging it as a false positive), mark it as reviewed, or escalate it. All actions are timestamped.
School counselor role
Gregarious does not have a direct integration with school counselors in the current version. The teacher is the first responder in the platform.
When a teacher escalates a concern to a school counselor, the teacher can share the relevant content and flag details directly. Future versions of Gregarious may include a counselor access role with appropriate permissions.
School counselors are the professionals best equipped to assess and respond to student welfare concerns. The platform supports their work; it does not replace it.
What parents can do
Parents linked to a student account can see their child's Studio projects and Pulse check-ins. If a parent notices something concerning in their child's writing, they should:
Parents can also disable their child's Pulse check-in feature from the parent portal settings. This removes the daily mood check-in from the student's experience.
Life safety resources
These resources are available 24 hours a day, 7 days a week. They are free and confidential.
988 Suicide and Crisis Lifeline
Call or text 988
Free, confidential support for people in suicidal crisis or emotional distress. Available 24/7. Trained counselors answer calls and texts. Also available at 988lifeline.org.
Crisis Text Line
Text HOME to 741741
Free, 24/7 crisis support via text message. Trained crisis counselors respond within minutes. Confidential. Available across the United States.
Childhelp National Child Abuse Hotline
1-800-422-4453
24/7 crisis intervention and support. Serves children, parents, and professionals. Provides referrals to local resources and emergency services. Calls are confidential.
Honest limitations
The AI detection system is not infallible
Automated detection of concerning content is a supplement to human attention, not a replacement for it. The system will produce false positives (flags on harmless content) and false negatives (missed genuine concerns). Teachers and parents should not rely on the platform to catch everything. Knowing your students remains the most important safety tool available.
If you have a specific concern about a student, contact them and their family directly. Do not wait for a platform notification.
Questions about this protocol?
Schools and districts can request a walkthrough of the safety features and the notification workflow before deployment. Email with subject line "Safety Walkthrough Request."
cgb003@gmail.com