Three-Tier AI Governance Classification

Tier 1

Routine Automation

Lightweight

Characteristics

AI that automates internal administrative tasks with no direct citizen or customer impact.

Document summarization Scheduling optimization Data entry automation

Governance Approach

Department-level approval, standard data quality checks, periodic review.

Efficiency & Service Data Quality

Tier 2

Decision Support

Moderate

Characteristics

AI that informs human decisions affecting citizens, customers, or resource allocation.

Workload prioritization Permit review support Trend analysis

Governance Approach

Documented human-in-the-loop requirements, transparency about AI's role, data privacy impact review.

Data Privacy & Security Human Decision-Making Transparency

Tier 3

High-Impact Decisions

Robust

Characteristics

AI that directly affects citizen or customer outcomes, rights, or access to services.

Eligibility determinations Risk assessments Automated decision comms

Governance Approach

Human oversight at the decision point, transparency to affected citizens, data quality validation, and a mechanism for concerns about automated decisions.

All Five Governance Pillars Apply

Why this matters: When Tier 1 use cases can move quickly under lightweight controls, teams are less likely to bypass governance altogether. That is one of the most effective ways to reduce shadow AI. Governance becomes something people work within — not something they route around.

Provenance: Developed through global comparative governance research at the Center for AI and Digital Policy (CAIDP). Part of the MPBP Framework for operationalizing AI governance.