Each test states a requirement, the evidence an assessor examines, the test itself, and pass and fail conditions. The first test, for example, requires an inventory of every actor capable of exercising informational power, including supervisors, administrators, developers, contractors, vendors, AI agents and service accounts. Omitting a materially capable actor or machine identity fails.
What the tests cover
| Area | Examples |
|---|---|
| Authority and purpose | Actor enumeration, information domains, originating purpose, purpose particularity, independent authority, authority-capability correspondence |
| Structural capability | Purpose-bound access, standing capability, government golden record, search boundary, material expansion gate |
| Correlation, inference and AI | Correlation and inference authorization, AI boundary inheritance, model materiality, composite machine action, feature authority, signal retention |
| Discovery and investigation | Discovery origin, objective predicates, discovery-to-investigation transition, retrospective surveillance, federated law-enforcement search, fraud-risk purpose |
| Intermediaries and brokers | Intermediary classification, broker disclosure and acquisition, commercial availability, reacquisition loops |
| Minimization | Verification alternative, question particularity, population-scale generality, audit separation |