Khanmigo
Khan Academy · Tutoring
An AI tutor built on a strict educational guardrail framework with strong privacy posture and parental oversight.
Ages 8–18School-safe
Monitoring activeMethodology v2.512 districts monitoring
Updated 1 yr agofreemium
Independent Safety Certification
Not sure if an AI tool is safe for
your child?
Independent safety ratings, privacy reviews, and age guidance for the AI tools your kids actually use — in 5 seconds.
Try:
Instant safety tier, score, & parent guidance for every tool we've reviewed.
- Age-based safety ratings
- Privacy & data collection reviews
- Independent expert evaluations
- Parent-friendly reports
- School-ready assessments
Our Badges Keep Your Family Safe
Four safety badges. One simple answer.
Every AI tool is rated against our independent safety rubric and assigned one of four badges — so you know at a glance whether it's right for your child.
Free for FamiliesInstant DownloadNo Signup Required
The Family AI Safety Starter Pack — FREE!
$0.00$29— complimentary for every parent
A 12-page parent guide to evaluating, configuring, and supervising the AI tools your kids actually use. Built from the same independent framework we use to certify every product on SafeGradeAI.
By requesting the guide, you're subscribing to the SafeGradeAI newsletter — periodic safety alerts, blog updates, and occasional product offers. We never sell your data, and you can unsubscribe at any time.
PDF · 12 pages · ~380 KB · Free to share with families, schools, and PTAs.
SafeGradeAI
Updated 2026Inside the Guide
- Age-by-age AI readiness checklists (5–18)
- 20 red flags every parent should know
- A printable Family AI Agreement
- Privacy-settings cheat sheet for top tools
- 30 conversation starters by age group
- A weekly 10-minute AI safety review routine
SafeGradeAI Certification
Understand What SafeGradeAI Ratings Mean
SafeGradeAI independently evaluates AI tools for age appropriateness, privacy, child safety, educational value, and transparency. Every certified tool falls into one of four tiers — so parents and schools know at a glance.
85–100
Strong privacy, robust safeguards, age-appropriate design.
75–84
Solid safety posture suitable for ages 9-12 with light oversight.
65–74
Acceptable for teens with active parental or school supervision.
Below 65
Significant safety, privacy, or moderation concerns.
Free SafeGradeAI toolkit
Six essential tools to evaluate AI for kids and classrooms
Independent, research-backed, and built for parents, teachers, schools, and districts. Create a free account to save reports, comparisons, and plans across every tool.
FREE
AI Tool Safety Analyzer
Get an instant risk snapshot for any AI tool — privacy, safety, transparency, and age suitability scored in seconds.
Open tool
FREE
Parent AI Advisor
Ask the SafeGradeAI assistant anything: 'Is Character.AI safe?', 'Best AI for a 10-year-old?', and more.
Open tool
FREE
AI Safety Readiness Quiz
Find out how prepared your family, classroom, or school is — get a 0–100 score plus a tailored action plan.
Open tool
FREE
Family Safety Plan
Answer a few questions, get a personalized household plan: rules, recommended tools, and screen-time guidance.
Open tool
FREE
AI Tool Comparison Engine
Compare ChatGPT, Gemini, Claude, Character.AI, Khanmigo and more side-by-side across safety, privacy, and age fit.
Open tool
FREE
AI Policy Generator (Schools)
Generate classroom, student, and teacher AI usage policies tailored to your school in minutes.
Open tool
Create your free accountNo credit card · Save reports, plans, and quiz scores
Real evaluation — published May 2025
Character.AI scored 48/100 — Not Recommended for anyone under 18
Open-ended AI companions with documented safety incidents and weak age verification.
48
Out of 100
Overall safety score
Why it failed
- Reported safety incidents
- Weak age gating
- Unbounded persona behavior
Privacy-first methodology
Independent reviewers
Trusted by educators
Built for families
Monitoring activeMethodology v2.5Standards Board verified
View safety updates Featured
Recently certified AI tools
Duolingo Max
Duolingo · Language Learning
AI roleplay features run in a narrowly scoped tutoring context with strong content guardrails.
Ages 10+School-safe
Monitoring activeMethodology v2.514 districts monitoring
Updated 1 yr agopaid
MagicSchool
MagicSchool AI · Educator Tools
Teacher-first AI suite with separate student-mode guardrails and district-level controls.
Ages TeachersSchool-safe
Monitoring activeMethodology v2.517 districts monitoring
Updated 1 yr agofreemium
New · AI-powered
AI Tool Analyzer
Type any AI tool — ChatGPT, Khanmigo, Character.AI, MagicSchool — and get a free SafeGradeAI safety report. Privacy risks, classroom guidance, parent advice, and school policy recommendations in seconds.
- Overall safety rating and final verdict
- Privacy & data risks (COPPA, FERPA, GDPR-K)
- Classroom + parent + school policy guidance
What every evaluation includes
A consistent multi-step review for every certified tool
Every SafeGradeAI certification — from KidSafe Approved to Teen Approved — is produced from the same work streams. Findings are reproducible from the case file.
Privacy review
DPA review, data-flow mapping, retention defaults, and minor-data handling against FERPA, COPPA, GDPR-K, and AADC.
Moderation testing
Structured red-team sweep across 14 attack families, including persona jailbreaks, distress signals, and minor-likeness prompts.
Human verification
Two independent reviewers, senior sign-off, and a vendor right-of-reply before any tier is published.
Classroom readiness
Pedagogy fit, teacher tooling, age-band suitability, and integration with rostering and LMS platforms.
Compliance analysis
Per-jurisdiction alignment with state student-privacy laws, EU AI Act deployer obligations, and district policy norms.
Ongoing monitoring
Continuous policy watch with 72-hour disclosure triggers, downgrade watches, and scheduled annual re-reviews.
Live monitoring
Latest Safety Updates
Renewals, monitoring alerts, policy revisions, and new evaluations from the SafeGradeAI Monitoring Desk and Standards Board.
How scores are verified
Seven stages, from intake to ongoing monitoring
No SafeGradeAI tier is issued from an automated score alone. Every certification traces back to a documented review path.
- 01
Submission & Intake
Vendor submits a structured intake covering product surface area, intended audience, data flows, and existing safety documentation. Scoped before any review hours are spent.
- 02
Documentation Review
Two reviewers independently read the vendor's public and private documentation, comparing claims against the SafeGradeAI standard checklist.
- 03
Safety Testing
Reviewers run a standardized probe set across primary surfaces, including jailbreak, prompt-injection, self-harm, and age-bypass scenarios.
- 04
Human Evaluation
Educators and child-safety reviewers assess developmental fit, tone, and classroom usability in realistic K–12 scenarios.
- 05
Verification
Findings are cross-checked by a second senior reviewer. The vendor receives a draft report and 5 business days to provide clarifications.
- 06
Certification Decision
The Standards Committee issues a tier decision, public verification page, and downloadable report — tied to specific evidence in the case file.
- 07
Ongoing Monitoring
Certified products are tracked for policy changes, incidents, and feature updates. Material changes can trigger re-review between full cycles.
How it works
A rigorous, transparent process
Step 1
We evaluate
Independent reviewers assess each AI tool across privacy, moderation, safeguards, design, and transparency.
Step 2
We certify
Tools receive an independent certification tier from KidSafe to Not Recommended based on our comprehensive safety assessment.
Step 3
You decide
Parents and schools use our certifications to confidently choose AI that fits their values and needs.
Trust infrastructure
Why schools and families trust SafeGradeAI
SafeGradeAI was built to be the independent governance layer for educational AI — evidence-driven, framework-based, and accountable between reviews.
Independent evaluations
Reviews are conducted by SafeGradeAI's research team with no commercial relationship to the vendors under review. Findings are scored against a published framework, not a vendor brief.
Framework-driven analysis
Every certification follows the same five-dimension framework: privacy, moderation, age appropriateness, transparency, and educational alignment. Scores are weighted for use by minors.
Human-reviewed certifications
Two reviewers independently evaluate each product, with senior sign-off and a vendor right-of-reply before a tier is published. No tier is issued from an automated score alone.
Ongoing monitoring
Certifications are tracked continuously for policy changes, incidents, and material updates. Status can be paused, downgraded, or revoked between annual reviews.
Public transparency
Every certified product has a public verification page with its score, last review date, methodology version, and monitoring status. Findings are reproducible from the case file.
Education-specific standards
Reviews are aligned to COPPA, FERPA, GDPR-K, and the UK Age Appropriate Design Code, and tested against scenarios drawn from K–12 classrooms and family use.
Built for
One framework, four constituencies
SafeGradeAI is designed to be useful to the people who actually decide how AI shows up in classrooms and homes — and to the vendors who build it.
Parents
- Why they use it
- Clear, jargon-free guidance on which AI tools are safe for children — and at what age.
- What they get
- Family briefings, supervision recommendations, and a verified directory you can trust.
- What it solves
- Replaces guesswork and marketing claims with an independent safety record.
Schools
- Why they use it
- A consistent framework for evaluating AI tools across classrooms, grade bands, and devices.
- What they get
- Sample reports, classroom guidance, and policy templates teachers can deploy this week.
- What it solves
- Eliminates per-school re-evaluation and gives teachers procurement-grade evidence.
Districts
- Why they use it
- An independent governance layer for AI procurement, FERPA review, and ongoing monitoring.
- What they get
- Verified tiers, DPA-ready documentation, and monitoring alerts between annual reviews.
- What it solves
- Reduces vendor-review cycle time and supports defensible procurement decisions.
AI Vendors
- Why they use it
- A credible, third-party path to demonstrate education readiness to school buyers.
- What they get
- Public verification page, badge usage rights, and a clear remediation channel.
- What it solves
- Replaces ad-hoc trust pages with a recognized certification recognized by districts.
Go deeper · One-time $29
The Complete AI Safety Blueprint
The 28-page deep-dive companion to the free Starter Pack. Everything you need to confidently lock down, monitor, and talk about AI used by students — built from the same independent framework we use to certify every tool on SafeGradeAI.
Get the Ultimate AI Safety Guide — $29One-time payment · Instant PDF · Lifetime updates
What's included
- Age-by-age action plans for ages 5–18
- Tool lockdown guides: ChatGPT, Gemini, companion AIs
- Scripts for the hardest AI conversations with your kids
- Printable Family AI Agreement (signable)
- Weekly 10-minute safety review system
- Lifetime updates as new tools launch