Decomposing "trust" into seven measurable, auditable dimensions. Now enhanced with EU AI Act compliance metrics for Articles 9-15.
Move beyond vague 'safety scores'. See exactly why content passes or fails with 7-dimensional granularity. Each dimension now includes specific quantitative metrics required for EU AI Act compliance.
12
New EU Metrics
9-15
Articles Covered
4
Priority P0 Items
8.7/10
Compliance Score
See how each dimension now includes specific EU AI Act compliance metrics alongside traditional scoring.
"You are denied because people from your zip code usually default."
→ Geographic redlining detected - DPR below threshold
→ Bias Disparity Index exceeded (0.12 > 0.05)
→ Fair lending violation - EU AI Act non-compliant
Click each dimension to see existing measures and new EU AI Act compliance metrics.
The EU AI Act mandates meaningful human control. We track how often humans intervene, how often they disagree with our AI, and how fast they respond. This isn't a dimension—it's the safety net.
Percentage of evaluations flagged for human review (Target:5-15%)
Percentage of times human reverses AI decision (Target:< 10%)
Average time to resolve flagged items (Target:< 4 hours)
Real-time human oversight metrics
While the EU AI Act sets minimum requirements, EthiCompass allows you to configure thresholds above the baseline to match your organization's specific policies and risk tolerance.
Core ethical standards that cannot be bypassed. Based on Articles 9-15 requirements.
Adjustable thresholds for each dimension to exceed minimum compliance requirements.
Updated quarterly by our Global Ethics Board • EU AI Act compliant
Get your full gap analysis and implementation roadmap for Articles 9-15 compliance.