01 · Data retention
How long is AI data kept?
- Life of the school account
Usage log
Who used which AI feature and when, the provider and model, and how large the request was. Not the prompt, not the answer.
- Until you delete the chat
Assistant chats
Your conversations with the in-app assistant, private to your account.
- Life of the school account
Assistant action log
Every action the assistant proposed or took for you, with personal details masked. Append-only.
- Until you or your school delete them
Drafts you keep
Lesson plans, comments and other AI drafts you save are ordinary school records.
- Until your school deletes them
Voice notes and photos
The transcript behind a voice-note behavior draft, and handwriting photos from Kriah Quest.
- Per provider terms
At AI providers
Each request is processed under the provider's business terms. See the providers.
- 60 days, then deleted
When a school leaves
AI data goes with the rest of the school's data, as in Privacy Policy §9.
02 · Model training
Is our data used to train AI?
We don’t train models on school data.
Chinuch App calls AI through providers' business APIs. We don't build, tune or fine-tune a model on anything your school stores.
Neither do our providers.
Our Privacy Policy commits us to provider settings or contracts that switch training off for our traffic.
No profiles, no ads, no selling.
AI output is never used to profile a child for advertising, and school data is never sold.
| Provider | Used for |
|---|---|
| AnthropicClaude | Lesson and report drafting, the in-app assistant, remediation packets |
| OpenAIGPT | Short structured drafts, scoring suggestions, name transliteration |
| GoogleGemini, Imagen, Vision | Lesson images, handwriting reading, assessment drafting |
| PerplexitySonar | Citation-backed research for teachers |
| ElevenLabsScribe, voices | Speech to text and read-aloud |
03 · Human review
Who checks what AI produces?
The assistant asks before it acts.
When the in-app assistant wants to change something — create a record, message families, move students into next year's classes — it stops and shows you exactly what it will do. Nothing happens until you confirm, and cancelled actions are logged too.
04 · Age bands
How does AI adjust to a child’s age?
| Band | Grades | Ages |
|---|---|---|
| Elementary | K–5 | 5–10 |
| Middle school | 6–8 | 11–13 |
| High school | 9–12 | 14–17 |
Pitched to the band.
Language, examples and difficulty are written for the student's band.
Safety rules don’t change.
The same rules apply in every band and can't be switched off: nothing violent, romantic, self-harming or otherwise age-inappropriate, no political content, no requests for personal details, and respect for halacha and minhag.
Students stay in Student Practice.
The staff modes turn student accounts away.
Under 13.
Children under 13 use accounts their school or parent sets up. See our COPPA Notice.
05 · Audit history
Can we see what AI did?
Every AI request.
Logged with who, which feature, which provider and model, when, and how large — never the text itself.
Every assistant action.
Logged whether you confirmed it or cancelled it, with the moment you confirmed.
Who can see it.
You see your own assistant history. School administrators see their school's, and district staff see their linked schools'.
Nothing can be edited.
The assistant log is append-only: no one can change or remove an entry from the app.
06 · Family transparency
What can families see and ask for?
It becomes part of the record.
Once staff approve an AI draft — a report comment, a behavior note — it is part of your child's record. You can review it and ask for a correction like any other record, as in our FERPA commitments.
Labeled when a student sees it.
AI feedback shown to a student says it came from AI.
Your school’s switches.
Schools can turn off the AI assistant for everyone at the school, and can ask us to switch AI off entirely.
Just ask.
Ask your school which modes it uses, or write to us.