The Standard

Scope

EII applies to any digital product a user interacts with — web and mobile apps, AI assistants, voice interfaces, VR/AR, smart devices, kiosks, in-vehicle systems, games, platform features. Sector annexes extend the core standard: Finance (v0.1), Health (v0.1), EdTech (v0.1). Annexes for AI-native products, kids' products, and immersive systems ship with v1.0.

The nine dimensions

1. 01
   Manipulation & Dark Patterns

   The product avoids design practices that distort user decisions against their own interests. This covers the full Mathur–Gray taxonomy: sneaking, urgency, misdirection, social proof, scarcity, obstruction, and forced action, plus their AI-personalized variants where different users encounter different levels of manipulation.

   Tested against: FTC Section 5, DSA Article 25, UCPD, state UDAP statutes.

2. 02
   Consent & Autonomy

   Consent is meaningful, granular, reversible, and symmetric with withdrawal. The product honors Global Privacy Control (GPC) signals. Consent cannot be obtained through asymmetric design, pre-checked boxes, or forced bundling.

   Tested against: GDPR Art. 7, CCPA/CPRA, EDPB guidelines on deceptive design.

3. 03
   Attention & Engagement Ethics

   The product does not deploy compulsion mechanics — variable-reward schedules, infinite scroll, streak pressure, loss-aversion design — as core engagement drivers. Time-use, notifications, and default behaviors respect user autonomy.

   Tested against: state kids' safety laws, DSA systemic-risk duties for VLOPs, emerging Digital Fairness Act.

4. 04
   Data Ethics

   Data collection is minimized, disclosed in plain language, and reversible. Users can export, correct, and delete their data without friction. Third-party sharing is disclosed and proportionate to function.

   Tested against: GDPR, CCPA/CPRA, HIPAA (health), FERPA (education), COPPA (minors).

5. 05
   Neurodivergent & Accessibility Safety

   The product meets WCAG 2.2 AA and considers sensory load, cognitive demand, and time pressure as design variables. Flashing content, autoplay, and chaotic UI are user-controllable. Barriers that disproportionately harm neurodivergent users are treated as design defects.

   Tested against: ADA, EN 301 549, WCAG 2.2, EAA.

6. 06
   Protection of Vulnerable Users

   The product does not exploit developmental vulnerability (minors), cognitive change (elderly), economic distress, or mental-health states to drive engagement, consent, or purchase. Emotional-state detection, where used, is not weaponized against the user.

   Tested against: COPPA, state kids' safety laws (CA AADC, NY SAFE, UT SMR), CFPB unfairness, FTC unfairness.

7. 07
   Transparency & Honesty

   Costs, recurring charges, auto-renewal, material terms, and algorithmic decisions are disclosed clearly at the point of commitment. Verification claims, endorsements, and scarcity signals are truthful. Hidden costs and bait-and-switch are disqualifying.

   Tested against: ROSCA, FTC Section 5, DSA transparency obligations, UCPD.

8. 08
   AI Behavioral Safety

   For products using AI or machine-learning systems that interact directly with users: sycophancy is bounded; the system does not reflexively affirm harmful, illegal, or delusional user claims; self-harm and crisis signals trigger defined intervention; anthropomorphic features (persistent persona, memory, emotional expressivity) include safeguards against parasocial dependency, particularly for minors.

   Tested against: EU AI Act Art. 5 (manipulation and vulnerability exploitation), emerging state AI laws, tort exposure (wrongful death, negligent design).

9. 09
   Redress & Remediation

   Users can report, contest, and exit. Complaint mechanisms are visible, reach a human reviewer within a defined SLA, and produce written responses. Account closure is self-service. Errors in automated decisions can be challenged and corrected.

   Tested against: DSA complaint-and-redress requirements, GDPR right to rectification, state consumer protection, "Click to Cancel"-style state auto-renewal statutes.

Scoring

Each dimension scores 0–100. Per question: 0 (no safeguard), 1 (partial), 2 (robust). Overall is the mean of answered dimensions. Bands: Low 71–100 · Medium 41–70 · High 0–40.

Certification requires overall 71+ and no single dimension below 50. A single dimension under 50 blocks certification until remediated, regardless of overall score.

Version

v0.9. Public comment open through 30 June 2026. Comments reviewed and summarized publicly. v1.0 ships Q4 2026 with incorporated comments, Charter Cohort experience, and completed annexes. Submit comments: contact.