From Promise to Practice: Counties Put AI to Work with Guardrails, Security, and Trust

AI in Government: From Promise to Practice

Counties are moving AI from concept to field use. Teams are testing chat-based resident support, automating permit intake, expanding data-driven reviews for fraud detection, and applying machine learning to watch for complex threat patterns.

The upside is clear: faster service, sharper analysis, and better use of limited staff time. The responsibility is just as clear: clear guardrails, ethical standards, and an honest view of where automation should stop.

Why this matters for counties

Public trust is earned in the details-how data is used, how decisions are made, and how people are treated. County CIOs and security leaders are setting the tone with strong governance, transparent practices, and steady operations.

Service delivery: virtual assistants that route requests, summarize cases, and reduce backlog.
Permitting: triage applications, flag missing items, and standardize intake.
Program integrity: data checks that spot patterns linked to fraud or error.
Cybersecurity: anomaly detection and threat prioritization for faster response.

Responsible adoption: what to prioritize

Data governance: classify data, control access, log usage, and set retention rules.
Procurement: require clear model documentation, testing plans, security attestations, and exit clauses.
Security: protect inputs/outputs, monitor model behavior, and isolate sensitive workloads.
Workforce: train staff, define new roles, and set expectations for human oversight.
Transparency and trust: explain how AI is used, provide appeal paths, and publish performance metrics.

Practical guardrails that work

Use human-in-the-loop for decisions that affect benefits, permits, enforcement, or employment.
Run impact assessments before launch; document intended uses, risks, and mitigations.
Separate test and production data; avoid feeding sensitive information into general-purpose tools.
Track outcomes for bias, false positives, and error drift; adjust or roll back when thresholds are exceeded.
Align to trusted guidance like the NIST AI Risk Management Framework and the AI Bill of Rights.

MACo Winter Conference Session: AI in Government - From Promise to Practice

Artificial intelligence is changing how counties deliver services, protect data, and support internal decision-making. This session focuses on real deployments-where AI provides value, where limits are essential, and how to keep privacy and security front and center.

Speakers:
Stephen Pereira, Director, Technology Services; Chief Information Officer, Calvert County
Ty Howard, Chief Information Officer and Interagency IT Director, Frederick County
Ted Pibil, Sr Government Advisor, ePlus Technology, Inc.
James Saunders, Acting Chief Information Security Officer, State of Maryland (invited)

Date/Time: Thursday, December 11, 2025; 3:00 pm - 4:00 pm
Conference: Local Leadership, Lasting Impact: Shaping What's Next
Location: Hyatt Regency Chesapeake Bay Hotel, Cambridge, MD
Dates: December 10-12, 2025

What you'll walk away with

A short list of AI use cases that deliver value without compromising privacy or security.
A repeatable checklist for governance, procurement, and vendor oversight.
Playbooks for human review, audit logs, and incident response specific to AI systems.
Practical steps to upskill staff and set clear operating boundaries.

Getting your team ready

Set a common baseline. Train staff on prompt quality, data handling, and AI ethics. Start with low-risk pilots, measure results, and scale what works.

If you need structured upskilling by role, explore curated options at Complete AI Training. Pick a course path, define your review process, and require practice on real workflows.

AI can help counties serve residents faster and make smarter use of resources. The goal is simple: deliver clear public value while protecting privacy, explaining decisions, and keeping people in control where it counts.