Like any new technology, the introduction of AI chatbots just over two years ago has led to varying approaches to adoption among local governments. Today, many municipalities have transitioned from cautious, and in some cases, enthusiastic experimentation, to fully integrated AI-enabled applications in their daily workflows. Understanding where you are on this journey and how to proceed thoughtfully is critical to successful implementation.

How Municipalities Are Using AI

Municipalities are typically using AI in four ways, often simultaneously:

Formal policy use —Authorized use by specific individuals or offices. These policies require engagement with tech managers to manage costs, integrate with existing systems, and ensure sound security practices. Proper governance ensures the responsible allocation of resources while effectively managing risks.

Shadow IT —When employees use unsanctioned technology without management approval. This is likely the most common approach today, especially with widely available free chatbot services like ChatGPT, Microsoft Copilot, or Google Gemini. While these tools can boost productivity, they create security vulnerabilities and data management concerns.

Locally built applications —When tech-savvy employees develop AI programming skills (aka “vibe programming”) to create chatbot-based applications. These tools can be used to complete forms, prepare documents, or respond to public inquiries, thereby enhancing productivity.

Integrated applications —Existing and new vendor products incorporating AI capabilities, from video and GIS applications to customer service chatbots. Vendors are rapidly embedding AI into their products, often with limited transparency regarding how these systems make decisions or process information.

Addressing Each Approach

Formal policy: Develop comprehensive policies covering data privacy, security protocols, and appropriate use guidelines. Review past Tech Updates columns from PM (December 2024, February 2025, and August 2025) for detailed perspectives on these issues.

Shadow IT: This is a tricky issue—acknowledging it obligates you to ensure its appropriate use. Consider developing guidelines for safe AI use that strike a balance between innovation and security. Creating channels for employees to suggest uses and tools can help bring shadow IT into the light.

Locally built applications: These can be highly engaging but must operate openly with carefully considered policies. Implement monitoring functions to track usage and maintain control over it. Creating an innovation sandbox with appropriate guardrails fosters creativity while managing risks.

Integrated applications: Vendor materials often suggest that AI solves everything and must be adopted immediately. Don’t believe the hype. While AI will affect many aspects of government administration, thoughtful implementation beats rapid adoption. Review existing policies to determine which ones might be affected when vendors add AI capabilities, particularly those related to internal controls in finance or data privacy in public safety.

Evaluating New AI Applications

When considering new applications, vendor claims are often sweeping and ambitious. Sound diligence includes:

• Having your tech team review infrastructure compatibility, security, bandwidth requirements, data management practices, and alternatives.

• Talking to peer governments that have deployed the product.

• Checking competing products to ensure you’re selecting the best solution.

• Understanding workflow implications and effects on employee training and responsibilities.

• Reviewing union contracts to determine if changes require negotiations.

• Projecting comprehensive costs and determining appropriate purchasing methods.

• Ensuring vendor contracts allow data export if you switch providers later.

Critical Implementation Considerations

Budget and Resources

While free chatbots serve as an entry point, AI incurs costs both upfront and in the long term. Initial costs often include licenses, hardware upgrades, consultant fees, and training. Ongoing costs may consist of subscriptions, maintenance, and updates. Start with small pilot projects to demonstrate value before requesting larger investments. Remember that savings often take time as people adjust to new workflows.

Staff Training

Even excellent AI tools fail without proper training and careful timing. Use a step-by-step approach rather than sudden workflow changes. Identify staff members who are interested in becoming champions and let them help others. Be clear that AI should augment your staff, not replace them. Getting employees involved early reduces resistance. Consider varying comfort levels with technology and create multiple training options to accommodate different learning styles.

Accessibility

AI systems must serve all community members fairly. When deploying resident-facing AI, such as chatbots, ensure they are accessible to people with disabilities and non-English speakers. Test with diverse groups before full launch. If your AI makes decisions that affect residents, understand how it was trained and monitor the results to ensure that no groups are disadvantaged. Check with vendors to determine if their systems were developed and trained with these issues in mind. Always maintain traditional service options for those without internet access or technological comfort.

Vendor Relationships

Before signing contracts, consider system compatibility with your existing infrastructure. Look for solutions using open standards rather than proprietary formats. Understand data security and appropriate use policies. Have your attorney review contracts carefully, particularly data storage policies, to ensure they meet public records disclosure and retention requirements. Understand the risks of potential vendor “lock-in” and how they can be managed.

Building Public Trust

Residents may worry about government AI use if there is no proper communication. Given the sensitivity of different use cases, consider involving the community early through public meetings, website updates, and council presentations. When AI is being trained or used to make recommendations, clearly explain what data you’re using, how decisions are made, and what human oversight exists. For significant applications, especially those involving facial recognition or surveillance, consider creating a citizen advisory panel. Transparency prevents misinformation and builds confidence.

Governance and Compliance

Larger organizations can consider establishing an AI committee with members from IT, legal, finance, and key service departments to develop policies and ensure consistent implementation. Establish clear rules for who can approve AI deployments and how they will be monitored and managed. Stay informed about rapidly changing state and federal regulations that affect AI use, particularly in areas such as privacy, transparency, and human oversight. Professional associations, such as ICMA, GFOA, and state municipal leagues, can help track new requirements.

Measuring Success

Have a plan to evaluate results, tracking time savings, cost reductions, error rates, staff satisfaction, and resident feedback. Be willing to adjust or discontinue systems that are not delivering the expected benefits. The best implementations include continuous improvement processes and periodic reviews to incorporate feedback and new capabilities.

Moving Forward

Be transparent about AI deployment within your community. With significant public attention on AI, transparency about its governmental use will prevent rumors and misinterpretation. For municipalities with tech advisory boards, this falls under their purview, providing an opportunity for citizen input and oversight.

Successful AI implementation requires balancing technology possibilities with organizational realities. Understanding the risks and rewards is critical. Avoid being swayed by promotional hype or vendor pressure. Learn from organizations with implementation experience.

AI is the latest in society’s technology evolution. Manage it well through thoughtful policies and implementation—don’t let it manage you. By carefully addressing these considerations, your municipality can effectively leverage the benefits of AI while avoiding common pitfalls. Remember that technology should support your community’s priorities, not dictate them. Always keep people—both staff and residents—at the center of your AI strategy.

Additional AI Resources for Local Government

These organizations provide free resources, training, and peer networking opportunities to help local governments navigate AI implementation. All three organizations offer resources specifically designed for the public sector, emphasizing responsible and ethical AI adoption that serves the public good.

InnovateUS

• No-cost, self-paced, and live training on AI, data, and digital skills for public sector professionals.
• Courses include “Using Generative AI at Work,” “Scaling AI in Your Organization,” and “Responsible AI for Public Professionals.”
• Curriculum created by and for government employees.
• Over 115,000 learners from 150+ government agencies.
• Interactive workshops on AI procurement, human-centered design, and ethical AI implementation.

Government AI Coalition (GovAI Coalition)

• Led by City of San José with 2,000+ public servants from 700+ local, county, and state governments.
• Free template policies, AI vendor questionnaires, procurement guidelines, and AI FactSheet tool.
• Cross-agency collaboration through regular meetings, working groups, and annual summit.
• Gives local governments a collective voice in shaping AI development.
• Tools aligned with emerging national standards.
• Helps agencies jumpstart AI governance programs without starting from scratch.

Apolitical

• Global learning platform serving 200,000+ public servants in 160 countries.
• Online courses, workshops, and peer communities on AI, climate, sustainability, equity, and governance.
• Courses co-developed with Oxford, the London School of Economics, and Georgetown.
• Government AI Campus with specialized AI learning resources.
• Access to best practices and global peer connections.
• Platform co-designed and funded by governments.

MARC PFEIFFER is a senior policy fellow and assistant director at Bloustein Local, a unit of the Center for Urban Policy Research at Rutgers University (marc.pfeiffer@rutgers.edu), assisted by Claude.ai.