High-risk without a FRIA. Blocked. Not warned. Blocked.
The EU AI Act mandates a risk classification framework with specific obligations for each tier. The Risk & Compliance Fabric implements this framework structurally — classification is automated, and enforcement at every tier is architectural, not advisory.
Art. 9 of the EU AI Act requires high-risk AI systems to implement a risk management system including a Fundamental Rights Impact Assessment. In most implementations, this is a documentation requirement — audited after the fact. In ORION, Gate 03 is an enforcement mechanism: if a high-risk system does not have a completed FRIA on record, execution is blocked. Not warned. Not flagged for review. Blocked.
Risk classification is assigned at system registration and continuously re-evaluated. The system enforces the specific obligations of each classification tier — automatically, on every request, without requiring operator intervention per execution. Human oversight is built into the architecture; it does not depend on human diligence.
Standard operation. Baseline enforcement pipeline active. No additional constraints applied.
Transparency obligations enforced. Output labeling and disclosure requirements applied per Art. 50.
FRIA gate active. Continuous monitoring, human oversight controls, and Art. 9 compliance enforced structurally.
Execution blocked at infrastructure level. No application-layer override exists.
High-risk systems require a completed Fundamental Rights Impact Assessment. Missing or expired FRIA = execution blocked at Gate 03, regardless of any operator action.
Risk level assigned at registration, continuously evaluated. Classification determines which enforcement gates are active for every request from that system.
Built natively for EU AI Act Art. 9, 10, 13, 14, 26, and 50. The regulatory framework is the architecture — not a layer added on top.
High-risk systems are evaluated on every execution. Risk state changes produce immediate governance response — not post-hoc review.
Technical details are disclosed only within controlled institutional engagements.