Automated Decisions

Version: 2026-05-09 Effective Date: 2026-05-09 Authoritative Privacy Policy: /legal/privacy-policy §6 and §10.3 This page provides the meaningful information required by CPRA (Cal. Civ. Code § 1798.185(a)(16)) and GDPR Article 22 / UK GDPR Article 22 about Koydo's automated decision-making — what it does, what signals it uses, what decisions it produces, and how to opt out.

Koydo uses an automated system called PRISM (Prismatic Learning Intelligence Signal Model) to personalize the learning experience. PRISM runs only when a user has enabled Koydo Intelligence ("KI") — KI is opt-in for all users, and off by default for children under 13. PRISM observes how a learner interacts with Koydo activities and adapts content difficulty, pacing, modality, and presentation in response. PRISM does not make decisions that produce legal effects (such as denying a service or opening / closing an account). Its decisions are pedagogical: which question comes next, how long until a break is suggested, whether to switch to a more visual or more auditory presentation, and so on. PRISM does not affect billing, account access, content moderation outcomes, or any external-facing decision. Content moderation and safety decisions are made by separate systems with human review escalation paths.

PRISM consumes 46 numeric measurements of how a learner interacts with activities. PRISM does not consume names, photos, audio recordings, or any user-generated content — only measurements of interaction patterns. The signal taxonomy is published in full in Privacy Policy §6.1 across seven categories: timing, touch, color preference, spatial preference, persistence and problem-solving, social interaction, and daily/session patterns.

PRISM combines the 46 signals into ten composite indices:

1. Cognitive load estimate — under-challenged / appropriate / overwhelmed
2. Emotional state estimate — thriving / neutral / struggling
3. Developmental stage markers — fine motor, spatial reasoning, abstraction
4. Learning modality preference — visual / kinesthetic / auditory / reading-writing
5. Engagement quality — actively engaged / passively present / disengaging
6. Frustration risk — real-time
7. Curiosity indicators — voluntary exploration
8. Confidence calibration — does confidence match performance?
9. Attention quality — sustained attention within a session
10. Intrinsic motivation — voluntary versus pressured engagement

These indices are stored as numeric estimates with timestamps. They are not used outside the Service.

The indices feed six pedagogical adaptations:

1. Adjust difficulty — make next question easier or harder
2. Switch modality — show more visual content if visual preference is high
3. Suggest breaks — when attention or engagement declines
4. Offer encouragement — when frustration risk is elevated
5. Optimize session length — recommend session duration based on observed attention patterns
6. Generate parent reports — provide parents with insights into learning patterns

PRISM data is never used for: advertising, behavioral profiling for non-educational purposes, sale to third parties, or any purpose outside the educational service.

The significance of PRISM's processing for an individual learner is the personalization of the learning experience inside Koydo: the difficulty, pacing, and modality of next content. The envisioned consequences are improved learning outcomes (the educational purpose) and improved engagement (the experience purpose). PRISM does not:

• Make any decision producing legal or similarly significant effects
• Affect price, access, eligibility, or refund decisions
• Share its outputs with third parties for any purpose
• Train any AI model on its outputs

You may opt out of PRISM-based automated decision-making at any time:

• Adults (18+): Account Settings → Privacy → Adaptive Learning → toggle "Use Koydo Intelligence" off.
• Teens (13–17): Account Settings → Privacy → Adaptive Learning → toggle off (with optional parental visibility if a parent dashboard is linked).
• Children under 13: Parents control this via the Parent Dashboard → Manage Child's Data → Adaptive Learning → toggle off. Opting in requires Verifiable Parental Consent and is off by default.

When PRISM is off, Koydo serves a non-personalized version of the adaptive learning experience: difficulty progresses on a fixed curriculum path, modality is not auto-switched, and break suggestions are time-based rather than signal-based. The Service remains fully usable. Opting out has no other consequences. There is no charge, no degradation of unrelated features, and no adverse effect on the user's account.

Under CPRA and GDPR / UK GDPR Article 22, you may request meaningful information about the logic involved in any specific automated decision and the consequences of that decision for you. To make such a request, contact privacy@koydo.app with subject "ADM Logic Request." Koydo will respond within 30 days, providing the information described in this page customized to the specific decision in question, plus any additional detail reasonably necessary to satisfy the request.

Although PRISM does not produce legal-effect decisions, you may at any time request a human review of any pedagogical adaptation that you believe is inappropriate. Contact privacy@koydo.app or use Account Settings → Privacy → "Request Human Review of an Adaptive Decision." A member of Koydo's pedagogical and engineering teams will review the underlying decision and respond within a reasonable timeframe, typically within 15 business days.

Automated Decisions Logic v2026-05-09 — Effective May 9, 2026 — koydo.app/legal/automated-decisions