A rating designation is an opinion—not a percentage.
AI governance rating scales should be categorical and bounded by explicit scope. A designation communicates governance maturity and evidence confidence for a defined system and date. It should not be read as a probability of safety, regulatory compliance percentage or credit rating.
Why analytical scores and ratings should stay separate
Analytical scores are useful for diagnostics, remediation prioritization and trend analysis. Rating designations are intended for executive interpretation and external decision contexts. Keeping them separate reduces false precision and helps preserve rating stability when small analytical movements do not change the underlying governance condition.
What can change a rating
Material system changes, incidents, evidence deterioration, unresolved critical findings, changes in intended use, significant vendor or model changes, and new jurisdictional obligations can all justify re-review. A rating should always carry a validity period or defined monitoring conditions.