Skip to main content
The Alpha Standard is the comprehensive framework used to assess AI governance across all rated companies. It has evolved through four versions as the field has matured.

​
Version History

​
Scoring Mechanics

Each dimension is scored 0-100 based on codified indicators. The composite score is a weighted sum based on the company’s Rating Profile (Developer, Deployer, or Hybrid).

​
Weights by Profile

All weight sets are validated to sum to exactly 1.0 in the scoring engine.

​
Cross-Dimension Checks

The Alpha Standard includes several cross-cutting mechanisms:
  • Concern Ceiling Logic: The independent concern screen can cap the overall grade at D, CCC, BB, or A regardless of the composite.
  • Critical Floor Checks: General floors apply across D1-D10, with added universal floors for D1 and D8, profile-specific floors for D2, D6, D9, and D10, and assessed-rating D12 floors for corrigibility.
  • Cross-Signal Inference: Signals detected in one dimension can affect scores in related dimensions. A cybersecurity incident (D9) that reveals poor incident response also impacts D2 (Safety) and D8 (Regulatory).
  • Committee Zone Coherence: High variance between dimensions within the same committee zone flags structural governance gaps.

​
Regulatory Alignment

Every dimension maps to specific regulatory requirements through the D2R Map (Dimension-to-Regulation Map):
  • EU AI Act articles and annexes
  • ISO 42001 clauses
  • NIST AI RMF functions and categories
The D2R Map is mandatory on every rating output, ensuring that governance scores translate directly into regulatory compliance posture.