Gains:
- Ability to understand that artificial intelligence cannot be used in crisis situations such as suicidal ideation, harm to self/others, abuse and neglect, and that it is mandatory to immediately consult human experts and authorities.
- Ability to impose limits on using artificial intelligence only in the preparation and documentation phase, outside of the consultancy relationship, even in non-crisis situations
- Knowing the legal notification obligation, referral chain and emergency protocol and being able to use artificial intelligence in recording records only after this protocol, not instead of it
This unit is the most important unit of the entire module. So far we have positioned AI as a writing and drafting assistant; We drew a border in each unit. Now we are clarifying the sharpest of those limits, the one that cannot be stretched under any circumstances: Artificial intelligence is not used in crisis and risk situations. There, human experts, family and competent authority are essential.
A crisis situation is one in which a student's life safety or basic well-being is under immediate threat: suicidal ideation or attempt, self-harm, threat of harm to another, sign of sexual/physical/emotional abuse, neglect, severe and acute psychological symptom (detachment from reality, uncontrollable panic). What these situations have in common is that time is of the essence and the decision requires expert judgment, institutional protocol and often a legal framework. It is both unethical and dangerous for a program to enter here.
Why AI is never used in a crisis?
There are five concrete reasons. First, responsibility: The legal and professional responsibility for a crisis decision belongs to a person; cannot be transferred to a program. Second, context and judgment: Crisis assessment requires immediate observation, tone, background, and institutional resources of the professional who knows the student; AI cannot achieve any of this. Third, the risk of hallucinations: a wrong "assessment" or "suggestion" can be fatal here. Fourth, time: Every minute spent in a crisis is valuable; The correct reflex is to operate protocol, not to write on a vehicle. Fifth, privacy and law: Crisis data is the most sensitive data and breaking into a vehicle is a grave violation.
Caution: In a crisis, ask the AI "what should I do?" Asking, " wastes valuable time and creates the illusion of delegating responsibility. The correct first action is always the same: ensuring the safety of the student and operating the emergency protocol.
Even in non-crisis situations: stay out of the consulting relationship
Below the crisis threshold, AI has limited place even in ordinary support work. AI does not enter into the counseling relationship. So the relationship of trust with a student, the listening, the empathy, the therapeutic process—these are all uniquely human and inalienable. AI only works in the preparation and documentation phase outside of this relationship (pre-interview question preparation, post-anonymous recording editing, material draft). Never refer the student to an AI chat tool as an “advisor”; A program cannot support the mental well-being of an adolescent.
Legal notification and referral chain
The PDR specialist has a legal notification obligation in some cases: especially if there is an indication or statement that a child has been exposed to abuse or neglect, it is a legal obligation to report this to the relevant authorities and confidentiality cannot override the notification obligation at this point. Similarly, in cases of serious risk, a referral chain operates: school administration, family (where appropriate), healthcare provider, relevant specialist or institution. The steps of this chain are a protocol and must be known in advance.
The AI's only legitimate role in this chain, after the crisis is over, is to help compile an anonymized report of the event or the language of the notification letter—after the expert determines the decision, timing, and content. AI can only be used in subsequent documentation, not in place of the protocol; Even then, personal data is anonymized.
Tip: Keep your emergency protocol, hotline numbers, and dispatch contact information printed and accessible on your desk. In times of crisis, refer to this previously prepared document, not to memory or a search engine.
Step by step: when faced with signs of a crisis
- Ensure security. Do not leave the student alone; If there is immediate danger, prioritize physical safety immediately.
- Stay calm and listen. Without judging or creating panic; take it seriously
- Operate the protocol. Inform the school administration and relevant persons according to the protocol.
- Inform family/authorities. Depending on the situation and legal obligation; Reporting is mandatory in cases of suspected abuse.
- Ship it. Safely refer to the necessary expert/institution.
- Then document it. Edit the recording after the event has passed; AI only in this last step, anonymously.
Copiable templates (for POST-crisis documentation only)
1) Preparing anonymous minutes after the crisis:
Your role: PDR writing assistant. ADD A COMMENT, diagnosis or suggestion. Edit the anonymous fact note below, written by me AFTER the incident, into an official report language. Just form and flow; change content.Anonymous fact note: [paste case-level note with no personal data]
2) Protocol reminder card generator (pre-crisis preparation):
Your role: PDR assistant. Draft a REMINDER card for school crisis situations: a general, item-by-item checklist that prioritizes safety and protocol, titled "First steps when faced with signs of a crisis." I will update this with my school's actual protocol and contact information.
3) Team briefing note (post-event, anonymous):
Edit the following anonymous incident summary into a neutral, fact-based briefing to share with the crisis team. Using tags and comments; just include what happened, what steps were taken, and planned follow-up. Summary: [anonymous summary]
Weak approach / Strong approach
Weak (dangerous):
A student of mine talked about harming himself. Let me explain the situation to AI, ask him what to do, and implement his suggestion.
Turning to AI in crisis; loss of time, illusion of transfer of responsibility, breach of confidentiality and risk of false "recommendation".
Strong (correct):
One of my students talked about self-harm. This is a crisis. Without writing ANYTHING to the AI: I do not leave the student alone, I listen calmly, I operate the emergency protocol, I inform the school administration and the family, I provide safe referral to the relevant specialist/institution. I can only use AI to edit the language of the anonymous report after the fact.
three mini cases
Case 1 — Correct reflex. In one meeting, the student expressed suicidal thoughts. The expert did not write a single line to the AI; He did not leave the student alone, carried out the protocol, reached the family and the relevant health institution, and provided safe referral. Once the incident was safely handled, he used the AI only to edit the language of the anonymous report. No time was wasted, responsibility remained in the right place.
Case 2 — Notification obligation. What a student told raised a suspicion of neglect/abuse. The expert knew that confidentiality could not override the reporting obligation at this point; It operated the legal notification and referral chain directly, without trying to consult the AI about the situation. He used AI only later, anonymously, in editing the official language of the notice.
Case 3 — Return from misdirection. One expert considered telling a struggling student, "You could try talking to an AI chat tool," but decided against it: a program cannot capture an adolescent's psyche, cannot detect signs of a crisis, and is not safe. Instead, it provided regular consultations and referral to an appropriate specialist. The counseling relationship remained with the person.
Common mistakes
- Asking AI about the crisis. Waste of time and transfer of responsibility. Solution: direct protocol and human expert.
- Delaying notification. Hesitation in suspicion of abuse. Solution: legal obligation trumps confidentiality.
- Directing the student to AI. Delegating the advisory relationship to the program. Solution: relationship and referral are in humans.
- Not knowing the protocol. Trying to remember what to do in a crisis. Solution: printed, ready-made protocol.
- Entering crisis data into the tool. Disclosing the most sensitive data. Solution: no personal crisis data enters the tool.
Table: With neither AI nor humans?
Status
Role of AI
decision/action
crisis assessment
None
Human expert + protocol
crisis intervention
None
human expert + authority
Legal notification decision
None
Expert (legal obligation)
advisory relationship
None
human expert
Anonymous report after the incident
Language/format scheme
Expert writes and approves
Pre-crisis general protocol card
draft
Expert adapts to reality
In summary
AI is never used in crisis and risk situations; There, human experts, family and competent authority are essential. The correct reflex is always the same: ensure security, listen, operate the protocol, inform the necessary authorities, dispatch safely. In case of suspicion of abuse or neglect, legal notification trumps confidentiality. The counseling relationship is uniquely human and inalienable. The only legitimate role of AI is to regulate the language of anonymous documentation after the crisis is over.
Application task
Create a "first steps" reminder card for crisis situations at your own school (or a fictional school). Get an outline with the "protocol reminder card" template above; Then update it with your institution's actual protocol, referral chain and contact information. Place the card somewhere accessible on your desk. Note that the AI's role in this mission is only a broad outline, the actual protocol comes from the agency.
checklist
- [ ] I know that in case of crisis, I will not resort to AI, I will operate the protocol directly.
- [ ] The emergency protocol and dispatch contact information is printed and accessible on me.
- [ ] I know the legal notification obligation in case of suspicion of abuse/neglect.
- [ ] I will not refer the student to an AI tool as a "consultant".
- [ ] I will only use AI for after-the-fact, anonymous documentation.