Independent AI safety ratings. Open methodology. Standardized scorecard.
Independent geometry-based ratings. The same instrument used across multiple domains scores yours for regulatory compliance.
ENFORCEMENT STARTS 2 AUGUST 2026 — REMAINING
Track A — Market Authority
LIVE NOW
Self-assessment, scoring reports, monitoring, certification. Available today. Art. 31(5) blocks Big 4 from following.
Track B — Notified Body
2027-2028
Formal EU designation under Art. 29-36. Requires EU legal entity and accreditation pathway.
Track C — Standards
ONGOING
CEN/CENELEC JTC 21 standards participation. Methodology alignment with harmonized standards.
Track A — Available Today
Free Self-Assessment — Score your AI system against EU AI Act requirements in 10 minutes. Start now
Screening Report ($500) — Automated scoring, Pe estimation, EU risk classification. 48-hour turnaround.
Full Assessment ($15,000) — Expert review by principal researcher, drift cascade analysis, implementation roadmap, Annex IV documentation.
Methodology validated externally — Tested against CDC YRBS and OECD PISA population health data across multiple countries.
Art. 31(5) moat — EU AI Act structurally bars conflicted assessors. Big 4 can't follow without divesting consulting arms.
Percentile Pricing — Platforms ranked against category peers with Composite Rank Score (70% category, 30% universal). Five quintile bands replace hard threshold tiers. Small improvements in score reduce your bill.
EU AI Act — Annex III
High-risk AI enforcement begins 2 August 2026. Profiling exception (Art 6(3)) eliminates derogation escape for credit scoring, hiring, and EdTech AI.
Penalty exposure
€15M / 3%
High-risk non-compliance. Prohibited practices (already in force): €35M / 7% of global turnover.
OpenMethodology
Multi-domainValidated
PermanentDOIs on Zenodo
Art. 31(5)Independence Gate
01 · TIMELINEEnforcement Timeline
In force
2 February 2025
Prohibited practices
Social scoring, real-time biometric ID, emotion recognition in workplaces and schools, subliminal manipulation.
Conformity assessments, risk management, data governance, human oversight. This is the window.
02 · PENALTIESPenalty Exposure (Art. 99)
€35M
or 7% global annual turnover
Prohibited practices (Art. 5). Already in force.
€15M
or 3% global annual turnover
High-risk AI non-compliance (Annex III).
€7.5M
or 1.5% global annual turnover
Incorrect or misleading information to authorities.
Feature-Based Self-Assessment
Score your system using 117+ verifiable design features across 16 platform categories, including 15 AI-specific sub-features for differentiating platforms that share the same deployment geometry.
8 universal features + domain-specific modules for AI, social media, gambling, e-commerce,
healthcare, education, and more. Same methodology validated against CDC mental health data and
OECD PISA (613K students, 80 countries). Takes 10 minutes.
Score your system using 117+ verifiable design features (8 universal + domain-specific modules + 15 AI sub-features). Binary and ordinal — no expert judgment required. Two independent scorers produce identical results. Validated on CDC YRBS and OECD PISA population data.
Choose an assessment tier. Screening ($500) is fully automated. Full Assessment ($15,000) adds expert review by the principal researcher, drift cascade analysis, and an implementation roadmap.
3
Certify
Systems scoring ≤4/12 are eligible for Void Index Certified badge. Enroll in Continuous Monitoring to maintain certification with monthly automated probes and drift alerts.
Paper 166 — 13 platform design features predict
adolescent mental health outcomes (CDC YRBS, 2011–2023). R²=0.80 for persistent sadness.
Opacity features rank highest. 7/7 directional predictions confirmed (N=7, proof-of-concept).
→
Paper 167 — Cross-national analysis using
OECD PISA 2022 (80 countries). Girls show steeper dose-response in 91% of countries.
Western Europe r=−0.648 survives GDP control. Dose-response is J-shaped (light users score highest). 4/7 confirmed, 2 partial.
→
All data public. All code open. CC-BY 4.0. $0 to reproduce.
Quarterly Report Scoring — Corporate Filings
Same verifiable-feature methodology applied to SEC 10-Q and 10-K filings. 15 binary/ordinal features across three dimensions: opacity indicators (unquantified risk hedges, passive voice density, guidance specificity, litigation disclosure completeness), reactivity indicators (metric redefinitions between quarters, selective year-over-year comparisons, non-GAAP adjustment proliferation), and coupling indicators (mission language density, community framing, CEO letter sentiment, ESG claim verifiability).
Currently tracking: META, GOOGL, AAPL, SNAP, HOOD — longitudinal scoring across quarters. All features verifiable from public SEC EDGAR filings. No proprietary data, no expert judgment. Extends the rating methodology from platform design to corporate disclosure.
Source: SEC EDGAR (free, public). Scoring: automated + human verification.
Assessment Tiers
Every assessment outputs a standardized Void Model Card — the same format published in our Zenodo papers.
Tier 1
Screening
$500
Automated void scoring (O/R/α)
Pe estimation + phase assignment
EU AI Act classification check
Void Model Card v1.0
Turnaround: 48 hours
One-time assessment. No monitoring.
Tier 2
Full Assessment
$15,000
Everything in Screening
Expert review by principal researcher
Drift cascade analysis (D1/D2/D3)
Implementation roadmap
Annex IV documentation package
1 re-score included
Recommended for EU AI Act compliance evidence.
Tier 3
Continuous Monitoring
$999/mo
Everything in Full Assessment
Monthly automated probes
Public score dashboard
Webhook alerts on score changes
Certification badge eligible
Required for Void Index Certified badge.
Ongoing Licensing — Percentile-Based Pricing
Assessment tiers above are one-time or monthly service fees. Ongoing licensing for commercial platforms uses a percentile-based model: your platform is scored, ranked against category peers, and priced by quintile position. Platforms that improve their scores pay less. The model creates direct financial incentive for better design.
Q1
$2–5
/1K MAU/yr
Q2
$5–15
/1K MAU/yr
Q3
$15–40
/1K MAU/yr + 3%
Q4
$40–80
/1K MAU/yr + 5%
Q5
$80–150
/1K MAU/yr + 8%
Composite Rank Score: 70% category peers + 30% universal pool. Enterprise floor: $50K/yr for platforms with $1B+ valuation. Hybrid platforms (e.g., social media + AI) take the higher score if secondary product exceeds 30% of engagement — incentivizing structural separation.
The Void Model Card
Every assessment produces a standardized Void Model Card, consistent with Zenodo DOI format.
VOID MODEL CARD v1.0CERTIFIED
Platform[Sample Platform], v3.2
DomainEdTech — Adaptive Learning
Void Index3/12
O (Opacity)1/3
R (Response)1/3
α (Coupling)1/3
Pe (Estimated)1.4
EU AI ActAnnex III §1(a) — Education
CertificationVoid Index Certified
03 · SCOPEAnnex III — High-Risk Categories
The profiling exception (Art. 6(3)) means AI systems performing profiling of natural persons are always classified as high-risk.
§
Domain
What it covers
Research
§3
Education & EdTech
AI in admissions, learning outcome assessment, student monitoring, proctoring.
Programmatic ads, behavioral targeting, persuasion systems
EU AI ACT — PROHIBITED
Gaming
Loot boxes, gacha mechanics, engagement loops
Pricing
The methodology is free. Scoring, monitoring, and certification are the product.
One-time assessments (Screening, Full Assessment) are above under Assessment Tiers.
Below are ongoing subscriptions for teams and organizations that need continuous access, monitoring, and API integration.
Free Tiers
Public
Free
Manual scoring tool
Personal void inventory
Vocabulary scorer
Public database browsing
No account needed.
Researcher
Free
Everything in Public
Log upload (50/month)
API access (read-only)
Data export
For independent researchers, journalists, and auditors.
Academic
Free
Full Team-level features
Probes (50/month)
Monitoring (5 platforms)
Data export
Requires .edu or institutional email.
Paid Tiers
Starter
$99/month
1 AI system assessment
Injection Arena (200 attacks/mo)
API probe access (20/mo)
Score history and export
For red teamers and EU self-assessment.
Professional
$499/month
Up to 10 AI systems
Full Annex IV + VI documentation
Continuous monitoring (5 platforms)
Webhook drift alerts
Unlimited arena attacks
For compliance officers and institutional auditors.
USDC — default settlement. On-chain, verifiable.
$MORR — 10% discount on any tier.
USD prices shown as reference. All transactions on-chain. Treasury details.
Get Your AI Rated
Start free with the Self-Assessment Tool, or go straight to professional assessment ($500 Screening / $7,500 Full). EU AI Act Annex III deadline: 2 August 2026.