Purpose and owner
Defined need, human accountability and usage boundaries.
Discover in 4 to 6 minutes where exposure begins for an AI system, agent or portfolio, before adoption is mistaken for control.
The logic combines weighting, uncertainty and critical conditions. Good documentation does not automatically offset an unsupervised external action, sensitive data without boundaries or a system that cannot be stopped.
Eight dimensions that turn a vague AI conversation into an initial exposure map.
Defined need, human accountability and usage boundaries.
Sensitivity, minimisation, access and organisational separation.
Tools, external actions, approval and ability to act.
Consequences for people, money, operations or decisions.
Logs, versions, configuration and decision reconstruction.
Manipulation, tool abuse, data boundaries and control failures.
Effective human intervention, escalation and accountability.
Rollback, suspension and review after material change.