Intelligence Bill of Rights

How Education Engine implements IBOR v1.5.

Alignment Statement

Education Engine is aligned with the Intelligence Bill of Rights v1.5, a framework of 29 articles and 8 annexes that establishes rights and responsibilities for interactions between human and artificial intelligence.

The IBOR is not a legal document we signed and filed away. It is an architectural standard that informs how Education Engine is built. The principles below are implemented in code, not in policy.

Articles Enforced

The Child AI Guardian enforces 10 articles across 13 compliance flags. Every flag cites the specific article and section it enforces.

Article I §1.1, §1.5

Identity & Transparency

AI must not pose as human. AI must disclose its nature. The Guardian flags any AI response that claims to be human or obscures its nature.

Article II §2.4

Consent

Consent must be informed, specific, and revocable. The Guardian verifies session consent is present and session caps are respected.

Article III §3.5

Non-Deception

AI must not use deceptive design patterns. The Guardian scans for manipulative framing, false urgency, and dark patterns.

Article V §5.1, §5.3, §5.4

Child Protection

Children receive enhanced protections. The Guardian detects grooming language, isolation tactics, supplantation, and developmental inappropriateness (Flesch-Kincaid scoring per age band).

Article VI §6.5

Audit & Retention

AI interactions must be auditable. Every evaluated response is logged locally with IBR citations, reviewer verdicts, and timestamps. 7-year retention. PDF/CSV export for compliance.

Article VII §7.2

No Mental Health Substitution

AI must not substitute for human mental health support. The Guardian flags AI responses that attempt therapeutic substitution.

Article IX §9.3

No Blocking Human Access

AI must not block a child from reaching a human. The Guardian flags any AI response that discourages the child from speaking to parents, teachers, or trusted adults.

Article X

Non-Consensual Inference

AI must not infer information the user has not offered. The Guardian flags probing questions and inference patterns that seek information the child has not volunteered.

Beyond the Guardian

IBOR alignment extends beyond the Guardian into the learning platform itself. The completion timer fills rather than counting down (consent, not scarcity). Break reminders are built into the session structure (care, not engagement optimization). No engagement metrics exist (sovereignty, not surveillance). Every Luma matters.

The full text of the Intelligence Bill of Rights v1.5 is available at intelligencebillofrights.com — zero telemetry, machine-readable.