AI Policy

AI Use & Data Policy

Last updated: June 1, 2026

This Policy explains how TaleTykes uses AI to support learning, what data may be sent to model providers, and how consent and human review protect child learners.

AI supports learning — it does not replace parents, teachers, or professional judgment.

AI outputs may be wrong or inappropriate. Always review before sharing, publishing, or using in reports.

Overview

This AI Use & Data Policy explains how Optiarms, Inc. dba TaleTykes uses artificial intelligence on TaleTykes, what data may be sent to AI providers, and what safeguards apply for families, child learners, teachers, and schools.

This Policy supplements our Privacy Policy and Terms of Service. Where a signed school or district contract provides more specific AI terms, that contract controls for that customer.

AI Features on the Platform

TaleTykes may use AI to support learning and administration, including:

Personalized story generation, illustration prompts, and reading comprehension questions.
Writing feedback, grammar suggestions, rubric scoring, and structure analysis.
Math hints, problem generation, misconception tagging, and adaptive coaching.
Assessment item generation, parent summaries, teacher summaries, and report drafting assistance.
Automated moderation assistance and safety classification.
Optional tutor-training capture for product improvement when explicitly enabled and consented.

AI Providers

AI requests may be routed to configured providers depending on environment, feature, and contract settings.

Anthropic (Claude) — primary provider for structured educational generation and scoring in production when configured.
xAI (Grok) — fallback provider for text and image generation when configured.
Deterministic mock providers — used in development and test environments only.
OpenAI — optional provider for oral-reading transcription (Whisper) when configured; separate from general text generation routing.
Future providers may be added and disclosed in our Subprocessor List.

What Data Is Sent to AI Providers

We send the minimum reasonably necessary prompt and context to perform the requested feature. Before external calls in production, learner and user identifiers are replaced with pseudonymous tokens, and certain sensitive patterns may be scrubbed from non-fiction student text.

Prompts may include instructional goals, reading level, draft text, answers, safety signals, and feature-specific context.
Story generation may retain hero names in narrative text for quality; other identifiers are pseudonymized.
We log provider, model, purpose, token counts, latency, estimated cost, and safety results for audit, billing, and operations.
Production use of Anthropic requires commercial terms confirmation (ANTHROPIC_COMMERCIAL_CONFIRMED) so student work is handled under appropriate enterprise agreements.

Model Training and Secondary Use

We do not use school-controlled education records to train our own general-purpose AI models unless a school contract expressly permits that use.

We do not authorize AI providers to use school-controlled education records or child personal information for targeted advertising.

Optional tutor-training capture is off by default in production configuration and requires explicit parent consent when enabled.

Safety, Moderation, and Human Review

Child-facing AI features use layered safety controls, including input moderation, prompt sanitization, schema validation, output moderation, image prompt review, logging, rate limiting, and human review workflows where appropriate.

Safety systems reduce risk but cannot guarantee every output will be correct or appropriate.
Adults must review AI output before instruction, publication, reports, or high-stakes decisions.
We may block, rate limit, flag, or route content for review when safety systems detect risk.

Limitations and No Guaranteed Outcomes

AI outputs are generated by probabilistic systems and may be incomplete, inaccurate, biased, outdated, or unsuitable for a particular learner or objective.

Mastery signals, recommendations, and AI-generated assessments are informational aids. They must not be the sole basis for high-stakes decisions about a learner.

Contact

AI and privacy questions: privacy@taletykes.com. Legal notices: legal@taletykes.com. Support: support@taletykes.com.

See also our Content Moderation Policy and Subprocessor List.