Scoring Methodology & Research Standards

An open, peer-reviewed evaluation framework grounded in international AI governance standards (NIST AI RMF, ISO/IEC 42001, EU AI Act).

1. Regulatory & Standards Foundation

Our risk scoring parameters are modeled directly after established global AI risk frameworks:

  • NIST AI Risk Management Framework (AI RMF 1.0): Informs our Governance (Govern) and Hallucination (Map/Measure) metric weighting.
  • EU Artificial Intelligence Act (2024): Provides the risk classification tier boundaries (Low, Moderate, High, Critical/Unacceptable).
  • ISO/IEC 42001 (Artificial Intelligence Management System): Informs our Human-in-the-Loop (HITL) and audit trail safeguard definitions.
  • GDPR Article 22 & CCPA: Dictates the mandatory data privacy weighting for automated profiling and customer data processing.

2. Mathematical Weighting & Scaling

Raw inputs are normalized on an integer scale from 0 to 100. Decimal rounding is performed using standard arithmetic half-up rounding (Math.round in client-side runtime).

3. Primary Source References

  1. NIST Special Publication 1270: Towards a Standard for Identifying and Managing Bias in Artificial Intelligence (2023).
  2. European Parliament: EU AI Act Text & Harmonised Governance Requirements (Official Journal of the European Union, 2024).
  3. ISO/IEC 42001:2023: Information technology — Artificial intelligence — Management system.

4. Limitations & Assumption Disclosures

This evaluation tool estimates static risk profiles for isolated workflows. It does not account for dynamic multi-agent feedback loops or active malicious adversarial prompt injection attacks. Organizations must conduct periodic security audits.