The Service uses artificial-intelligence models (from providers including Anthropic, OpenAI, Google, Perplexity, and ElevenLabs) to assist with tasks such as generating lesson materials, drafting parent communications, transcribing speech, grading open-ended student responses, and building remediation packets. By using the Service, you acknowledge and agree to the following:
14.1 AI output is a draft, not truth
AI-generated content may contain inaccuracies, biases, or errors. All AI-generated report-card narratives, parent emails, remediation plans, and grading suggestions must be reviewed by a qualified human (teacher, administrator, or parent) before being treated as authoritative. Chinuch App does not warrant that AI output is accurate, complete, pedagogically optimal, or suitable for any particular student.
14.2 AI disclosure to students
Content shown to a student that was produced by AI is visibly labeled as "AI-generated" and includes a way to report inappropriate output. Users under 18 interacting with AI-backed features will be informed that they are communicating with an AI assistant and not a human.
14.3 PII handling
Before any prompt leaves our servers, it passes through a centralized PII-scrub that removes or replaces a child's name, email, phone, street address, date of birth, NYC OSIS / school ID, parent contact information, and medical or financial identifiers. See Section 14 of our Privacy Policy and Section 6 of our COPPA Notice for full detail on how we sanitize outbound AI requests.
14.4 Child-safety policy
Every AI call that can reach a minor is wrapped with a non-negotiable child-safety preamble that prohibits sexual, violent, self-harm, substance-use, weapons, political, and age-inappropriate content, in line with the providers' own policies for organizations serving minors. We prohibit use of the Service to elicit or produce such content.
14.5 Prohibited uses of AI features
You may not use AI features to:
- Produce content targeting a specific child with harassment, shaming, or content that a reasonable educator would not produce.
- Impersonate a real person (teacher, parent, rabbi, clinician).
- Generate medical, psychological, legal, or safety advice presented as authoritative.
- Attempt to extract training data, model weights, or prompts from the underlying providers.
- Bypass or defeat the safety guardrails or PII scrubbing described above.
14.6 No training on customer data
Our AI subprocessors are contractually prevented from using your data, your school's data, or your students' data to train their models.
14.7 AI credit pools
Some AI features consume a per-school token allowance. Usage is logged per user and per school and is visible to school administrators. Exceeding the allowance may temporarily disable AI-backed features until the next billing cycle or until additional credits are purchased, without affecting the rest of the Service.
