Civic Governance Atlas

Map authority before recommending action.

The Atlas is NCI’s public governance framework for systems where authority, technology, vendors, data and institutional rules overlap. It maps purpose, decision rights, evidence, implementation, accountability and exit before treating a policy or technology as a solution.

What the Atlas doesMaps who decides, who pays, what evidence counts, who controls data and how a system can be challenged or exited.
What it is notA score that declares a technology or policy inherently good or bad.
Primary usePublic-sector decisions where authority and accountability are split across institutions.

AI & Automated Systems Governance

Nine core questions, three supplemental checks.

NCI separates innovation policy, institutional deployment rules and infrastructure governance. The public question is not whether a system is simply “AI”; it is how authority, data and accountability change when the system is used.

Every case answers the nine core questions below. Three supplemental checks sit under the Data, Evaluation and Accountability questions: sensitive-location risk, verification provenance and assumed user knowledge.

PurposeWhat public problem is the system meant to solve?
AuthorityWho can authorize, limit or stop its use?
ProcurementWhat evidence, contract terms and vendor conditions are required?
DataWhat data enter the system, who controls them and where do they go?Sensitive-location riskDoes the system collect data at or near shelters, clinics, schools, houses of worship or similar sites, where being tracked can carry greater consequences, and do its rules treat those locations differently?
DecisionIs the system advisory, or can it materially affect rights, benefits, discipline or access?
Human controlWho can override, reverse or halt an automated outcome?
EvaluationWhat must be tested before and after deployment?Verification provenanceHow were claims about the system verified, including where devices are located? Public maps and advocacy datasets should say how each entry was confirmed and separate confirmed devices from suspected ones.
AccountabilityWho receives complaints, audits performance and answers for failure?Assumed user knowledgeWhat does the system expect the affected person to understand in order to use it safely, and has the institution actually provided that information? This matters most when a system tells users to verify its answers themselves.
ExitHow are contracts ended, data deleted, vendors changed and service continued?
Research architecture

Procurement and exit are not separate Atlas modules. They are two of the nine core questions applied to every case. Each case also carries standard procurement details, exit conditions and a dated change timeline so NCI can show how a system changes after launch rather than freezing the record at purchase.

Surveillance & Sensor Governance · Live

License-plate readers are the first full case. Other sensor technologies join the comparison once the evidence base supports it.

Automated Decisions · Next

Public AI and administrative decision systems sit in one track. MiDAS is the leading first case because implementation, automated fraud adjudication, court challenges, settlement and replacement are already documented in public records.

Digital Infrastructure · Later

Data centers remain in scope as infrastructure governance and stay on the watch list. Michigan now has documented records, including the MPSC’s October 1, 2026 order on DTE’s contracts with Google (Case U-22058), and NCI logs them. A full case opens only when the evidence is repeatable and NCI can define the audience, show what the case adds and maintain it over time.

Source-checked Michigan baseline: DTMB's public AI guidance is expressly guidance rather than policy; external tools processing state proprietary data require an Authority to Operate through MiSAP, and software purchases remain subject to state procurement rules. MDE's current district resources emphasize purposeful use, privacy, human oversight, equity, transparency and continuous improvement. DTMB source → · MDE source →
MiDAS case file: Michigan's Auditor General documented MiDAS implementation and control weaknesses; the Attorney General later announced a $20 million settlement resolving a class action alleging false fraud accusations and property seizures without due process; UIA is now transitioning to MiUI, with benefits functions scheduled to launch November 9, 2026. NCI will use the lifecycle as an Automated Decisions case rather than as a generalized claim about all AI systems. OAG audit → · AG settlement → · MiUI transition →

Michigan Data Center Governance Scorecard

Treat data centers as infrastructure governance.

The scorecard in development will examine who decides, who pays, what must be disclosed and how communities or regulators can review outcomes after a project is approved.

Authority and siting

State, utility and local decision rights; land-use authority; permits; incentives; and the points where public input or appeal exists.

Infrastructure and cost

Generation, transmission, distribution, water and other infrastructure requirements, including which parties bear incremental costs.

Public claims and disclosure

Jobs, tax revenue, economic-development promises, resource use, timelines and the information needed to test those claims later.

Review and exit

Performance reporting, enforcement, changed conditions, closure obligations, transition risk and what happens if expected benefits do not materialize.

Current federal context: H.R. 9340, the Ratepayer Protection Act, passed the House 417-3 on September 16, 2026. On September 30, the Senate rejected cloture on the motion to proceed, 57-43; the measure did not advance through that procedural vote and is not enacted law. The proposal addresses recovery of full incremental utility-upgrade costs serving large-load customers. GovInfo bill record · September 30 Senate vote

Surveillance Governance Series

One governance method across changing surveillance tools.

The series expands NCI’s existing ALPR and gunshot-detection work into a coherent research franchise: license-plate readers, facial recognition, acoustic gunshot detection, drones, commercial/location data, connected vehicles and integrated vendor platforms.

Need and authority

Define the problem, legal authority, permitted uses and the decisions the system can support.

Data and supplier control

Map collection, retention, sharing, cybersecurity, subcontractors, model changes and access across jurisdictions.

Evaluation and exit

Measure operational and public outcomes, not only technical accuracy, and define renewal, termination, deletion and transition rules.

Federal development: H.R. 10366, the No Federal License-Plate Observation and Camera Keeping Act, was introduced September 14, 2026 and referred to the House Committee on Transportation and Infrastructure. The proposal would condition surface-transportation funding on state restrictions governing automated license-plate-reader use; it is not law. GovInfo bill record

Consumer data and automated markets

Track how data changes price, access and accountability.

NCI’s surveillance-pricing work examines how personal or inferred data can shape prices, fees, discounts and access to commercial offers, while distinguishing individualized data use from ordinary market-wide pricing.

Read The Price Should Not Know Who You Are

Research connected to the Atlas

Current briefs that test the framework in real systems.

The Atlas is not a separate shelf of theory. NCI uses it to organize research on procurement, surveillance, consumer data and institutional accountability.

Surveillance Governance SeriesProcurement

Buy Outcomes, Not Technology

A risk-based procurement framework for connected, data-intensive and AI-enabled public-safety systems.

Surveillance Governance SeriesGunshot detection

Beyond the Alert

A performance, evidence and contract framework for acoustic gunshot-detection technology.

Civic Governance AtlasConsumer data

The Price Should Not Know Who You Are

A Michigan framework for data-driven price discrimination, vendor responsibility and auditable pricing records.

Foundational archiveALPR governance

Automatic License Plate Readers and Municipal Data Governance

NCI’s earlier ALPR brief on retention, transparency, procurement, equity, access and data sharing.

Monitoring file

Election administration and public access.

NCI is monitoring election administration as a governance problem involving overlapping federal and state authority, operational capacity, civil liberties and institutional trust. Any public product will focus on who controls which decisions and how implementation works rather than campaign advocacy. Michigan’s next statewide election is November 3, 2026. Absentee ballots became available September 24; the standard statewide early-voting period is October 24-November 1; and the online/mail voter-registration deadline is October 19. Michigan election dates →