Conditions before use
Local-first decision support
Turn one proposed AI use into a reviewable plan.
This tool applies published pediatric, education, child-rights, privacy, and AI-risk guidance. It does not diagnose a child, certify a product, determine legal compliance, or replace review by a school, clinician, privacy officer, or counsel. Education, not medical advice.
Privacy: Nothing is sent to a server. Describe the use without entering a child’s name, diagnosis, school record, private conversation, or other identifying information.
Decision context
Who is using what, and why?
No child information is required. Record only what another adult needs to understand the proposed use.
Use analysis
What role will the system play?
The same model can create very different developmental and governance risks depending on its role.
Safeguard check
What must be true before use?
Answer for the tool as it is actually set up, using settings you can check yourself.
Decision record
A recommendation you can inspect, with every reason shown.
The result is conservative where evidence is limited and does not certify a vendor or resolve applicable law.
Review needed
Complete the review.
Questions to ask
Protect the developmental job
Monitor and revisit
Preserve the decision record
Export a plain-language packet for a co-parent, teacher, school review, product team, or qualified advisor.
Evidence desk
What this decision model relies on.
Sources are dated and linked so users can inspect the underlying guidance. Product behavior and rules can change; re-check before adoption. Dates and versions were checked on October 13, 2026.
Pediatrics (American Academy of Pediatrics journal) · Grundmeier et al., March 2026
“Generative Artificial Intelligence: Implications for Families and Pediatricians,” Pediatrics 157(4):e2025074912. This state-of-the-art review emphasizes developmental differences, potential benefits, misinformation and privacy risks, false perceptions of AI as a friend or caregiver, AI literacy, critical thinking, and safe integration.
Read the review →UNICEF · Guidance on AI and children, version 3.0, December 2025
Child-centered AI should support development and well-being, protect data and privacy, provide safety, fairness, transparency, explainability, accountability, inclusion, and child participation.
Read the guidance →UNESCO · Guidance for generative AI in education and research, 2023
Guidance calls for human-centered, age-appropriate use, data protection, institutional validation, teacher capacity, and careful regulation of untested tools.
Read the guidance →U.S. Department of Education · FERPA and online tools (FAQ; the page shows no date, so verify it is current)
Schools should confirm approval, direct control, authorized use, limited redisclosure, and consultation with administration and IT before tools receive education-record information.
Read the guidance →Federal Trade Commission · COPPA Rule amendments, finalized January 16, 2025
The amended rule strengthens parental choice, data minimization and retention limits, security requirements, and restrictions surrounding children’s data.
Read the update →NIST · Generative AI Profile (NIST AI 600-1), July 2024
A cross-sector risk-management resource for governing, mapping, measuring, and managing generative-AI risks across the system lifecycle. NIST’s web page for it was last updated in April 2026, but the profile itself is still the July 2024 edition.
Read the profile →The decision architecture
Six questions before “yes” or “no.”
01 · Development
What capacities and relationships are still being built?
02 · Purpose
What precise job is AI doing for the child or adult?
03 · Substitution
What thinking, practice, care, or connection might it replace?
04 · Data
What leaves the child’s control, who receives it, and for how long?
05 · Accountability
Which adult can inspect, intervene, correct, and stop the use?
06 · Evidence
What benefit is expected, what harm signals matter, and when will the decision be revisited?