Home › Criteria

Registry · v1

AI Posture criterion IRIs

https://aiposture.org/criteria/v1/

Stable test criterion IRIs for the AI Posture EARL v0 wire format. One IRI per vector per maturity level. Level 0 (N/A) is a falsifiable scope boundary defined in the specification, not a test criterion, and has no IRI.

These IRIs are stable. A criterion is never re-minted; a superseded criterion is replaced by a new IRI under a new namespace version, and the old IRI continues to resolve to its original meaning. The machine-readable registry is at index.json.

People

LevelNameAssertion
L1PerceivingOur people use AI. We have not measured how.
L2AssessingWe have looked at how our people collaborate with AI.
L3IntegratingWhat we saw changed what we do.
L4CalibratingWe measure continuously. Data drives governance.
L5EngineeringHuman-AI collaboration is an engineered capability, continuously maintained.

Infrastructure

LevelNameAssertion
L1PerceivingAgents may interact with our systems. We have not measured readiness.
L2AssessingWe have scanned our infrastructure for agent readiness.
L3IntegratingWe have acted on what the scan showed.
L4CalibratingWe scan continuously. Drift is tracked.
L5EngineeringAgent-readiness is engineered into how systems are built and shipped.

Regulation

LevelNameAssertion
L1PerceivingWe are exposed to AI regulation. We have not mapped it.
L2AssessingWe have mapped our obligations across jurisdictions.
L3IntegratingObligations are translated into controls.
L4CalibratingControls cover all in-scope jurisdictions. Regulatory change is monitored.
L5EngineeringCompliance is continuous. New jurisdictions onboard rapidly.

Use in EARL

Each criterion IRI is the earl:test of an earl:Assertion. See the profile and example for usage.