AI Learning Modules: AI Guardrails
University of Arizona |College of Education
Module two of a four-module program helping educators build foundational AI literacy, establish ethical guardrails, master prompt design, and navigate AI tools with confidence.
Created by Austin Ross, M.Ed., MA
Access U of A GenAI

Module Overview
This module guides University of Arizona educator preparation program instructors through the essential guardrails for responsible AI use in educational settings. Three core topics are covered in sequence.
01
Key Terms
Foundational vocabulary and concepts for understanding AI in education.
02
Human-in-the-Loop
Expectations and practices for keeping educators actively involved in AI-assisted decisions.
03
FERPA & Sensitive Data
Protecting student privacy and understanding data compliance when using AI tools.
Guardrails Welcome Video
Key Terms
Before diving into AI guardrails, it's essential to establish a shared vocabulary. Understanding these foundational terms will help educators navigate AI tools confidently and responsibly in their instructional practice.
AI Guardrails
Policies, practices, and boundaries that guide the safe and ethical use of artificial intelligence in educational environments.
Human-in-the-Loop (HITL)
A framework ensuring that human judgment remains central to AI-assisted processes — educators review, verify, and act on AI outputs.
Sensitive Data
Any information that could identify or harm a student if disclosed, including academic records protected under FERPA.
Defining FERPA and COPPA
FERPA, the Family Educational Rights and Privacy Act, protects the privacy of student education records, while COPPA, the Children’s Online Privacy Protection Act, regulates how websites and online services collect, use, and share personal information from children under 13.
Human-in-the-Loop (HITL)
What Is HITL?
Human-in-the-Loop means that educators remain active, informed participants in every AI-assisted decision — not passive recipients of automated outputs.
AI tools support instruction, but the educator's professional judgment is always the final authority.
HITL Expectations
A 5–10 minute recorded review of HITL expectations is included in this module. Key principles include:
  • Review AI-generated content before sharing with students
  • Verify accuracy and appropriateness of AI outputs
  • Maintain accountability for all instructional decisions
  • Document how AI tools are used in your practice
HITL in Practice
Keeping humans in the loop is not just a policy requirement — it's a professional responsibility. Educators must actively engage with AI outputs rather than simply accepting them.
Review
Examine every AI-generated suggestion, lesson plan, or assessment item before use.
Verify
Cross-check AI outputs for accuracy, bias, and alignment with learning objectives.
Refine
Modify AI-generated content to fit your students' specific needs and context.
Decide
Make the final call — your professional judgment supersedes any AI recommendation.
Human-in-the-Loop

FERPA & Sensitive Data
FERPA (Family Educational Rights and Privacy Act) governs how student data must be protected. When using AI tools, educators must ensure that no personally identifiable student information is entered into external platforms without proper authorization.
What FERPA Protects
Academic records, grades, disciplinary files, and any information that could identify a student.
AI Tool Risk
Entering student names, IDs, or performance data into unapproved AI platforms may violate FERPA.
Best Practice
Always anonymize student data before using AI tools, and only use district- or institution-approved platforms.
FERPA AND SENSITIVE DATA
Real-World Context: The RYG System
Teachers in the HCAT program have shared through post-observation conversations that their districts are utilizing a Red-Yellow-Green (RYG) light system to classify AI tool use by data sensitivity — as well as MagicSchool.AI. The RYG system will be explored in greater depth in Webpage 4 of this module series.

The RYG framework helps educators quickly identify which AI tools are approved (Green), require caution (Yellow), or are prohibited (Red) for use with student data.
RYG Classification Framework
Red: Prohibited
AI tools that cannot be used with student data. Examples: consumer AI tools (ChatGPT free, Google Bard personal accounts), any platform not reviewed by the district.
Yellow: Use with Caution
AI tools that may be used with anonymized or non-sensitive data only. Requires educator judgment and awareness of data risks.
Green: Approved
District- or institution-approved AI platforms (e.g., MagicSchool.ai, U of A GenAI) that meet FERPA and COPPA compliance standards.
FERPA Compliance in Action
Following this three-step process helps educators stay FERPA-compliant while still leveraging the power of AI tools in their instructional practice. When in doubt, consult your district's data privacy officer before entering any student information into an AI platform.
Why Magicschool.ai?
MagicSchool.ai is built with a commitment to security and privacy, making it a trusted platform for educators. Here is why it is safe to use:
1
FERPA-compliant: Ensures strict protection of student education records.
2
COPPA-compliant: Protects the privacy and safety of children under 13.
3
Data encryption: Utilizes advanced security measures to keep all information secure during transmission and storage.
4
No data sharing with third parties: User and student data is never sold or shared with outside organizations.
5
Designed for education: The platform is purpose-built to meet the specific safety and pedagogical needs of schools and teachers.
6
Transparent privacy policies: Maintains clear, accessible documentation regarding how data is collected, used, and protected.
Pause + D0
Choose a lesson plan or activity you used recently. Then, explore the available tools on Magicschool.ai and select one that could help you enhance or build upon your original plan.
Differentiated Reading
Formative Check
Engagement Hook
Text Scaffolder
Accommodation Suggestions
Remember, AI is your assistant, not your pedagogy.
Module Summary
This AI Guardrails module has introduced the foundational concepts educators need to use AI responsibly in University of Arizona preparation programs.
1
Key Terms
Shared vocabulary for AI guardrails, HITL, and sensitive data in educational contexts.
2
Human-in-the-Loop
Educators must review, verify, and take responsibility for all AI-assisted instructional decisions.
3
FERPA & Sensitive Data
Student data must be protected — use only approved tools and anonymize information before AI input.
4
RYG System
A practical district-level framework for classifying AI tool safety — explored further in Webpage 3.
End of Module
You have completed the AI Guardrails module of the University of Arizona Educator/Instructor Preparation Programs AI Learning Series.
Continue your learning with the next module: AI Prompting for Lesson Design.
Credits: Created with images by brent coulter — "Sonoran Sunset" • Jayeda akter — "HUMAN-IN-THE-LOOP isolated on Transparent Background"