Paragon Fellowship

Lebanon, NH

Proposed Revisions for ADM-143
and AI Usage Guidelines



Project Description

The City of Lebanon, New Hampshire has demonstrated a commitment to using artificial intelligence (AI) to enhance city services and productivity. In December of 2023, the City adopted its first AI policy “ADM-143: Use of Artificial Intelligence.” With the aim of transforming the broad strokes framework to a more comprehensive and actionable policy, the Paragon Summer 2025 Cohort has drafted proposed policy revisions to ADM-143 and a set of complementary AI guidelines for policy implementation. Our team sought to create a policy that ensures safe and ethical use of AI and adapts to the evolving AI landscape. To inform our work, the team conducted a policy evaluation, literature review, and stakeholder interviews.

Project Aims

In order to make Lebanon’s ADM-143 policy more robust and applicable to the various ways city employees use AI, our project focused on a few main goals:

1. Providing clear guidance to employees: We aimed to reduce the ambiguities present in ADM-143 version 1 by clarifying responsibility under the policy, providing clear instructions and example language for disclosing AI use, and specifying when certain sections of the policy apply to specific AI uses.
2. Expanding policy coverage: We included new considerations like environmental sustainability and restrictions on high-risk cases like realistic image generation.
3. Adapting to new technological developments: We updated multiple parts of the framework, such as risk review and human oversight, to consider the challenges posed by new technologies like AI agents. In some cases, we created distinct requirements for generative vs decision-making AI to address the unique concerns associated with each.
4. Distributing expertise citywide: We created a strong governance framework to reduce the burden on Lebanon’s Chief AI Officer and encourage more employees to actively promote responsible AI use. We clarified the responsibilities of the Technology Review Committee and department heads, and created a new “AI Champion” role within departments to help review technology use and disseminate information across the city. We also included recommendations on AI training to boost city capacity as a whole.

Methodology

In formulating our revisions to ADM-143 and the creation of the supplementary guidelines for implementation and use cases, we moved through three main stages that informed and influenced our decisions. Each stage was conducted with the intention of grounding our recommendations in reputable research while taking into account the specific needs of the City.

1. Evaluation of ADM-143: We began by reviewing the City’s first version of ADM-143, adopted in 2023, to understand where the policy was at and what direction we needed to go. We conducted a comparative policy analysis that examined the AI policies of six other municipalities that were strong, comprehensive, and relevant to Lebanon. This was crucial to our understanding of where Lebanon’s policy shined and where it didn’t and helped us identify gaps and areas for refinement in our policy revisions.
2. Literature Review: To serve as a knowledge base to inform our understanding of the municipal AI policy climate, we conducted preliminary research on a wide array of sources. Our literature review included scholarly research and peer-reviewed journal articles, government frameworks and guidelines, municipal case studies, and reports from non-governmental organizations and advocacy groups. Key findings from the literature review were used to guide the principles behind our policy and guidelines, prioritizing community engagement and inclusion, flexibility, and responsible experimentation.
3. Stakeholder Engagement: Given that ADM-143 and its supplementary guidelines directly impact the City of Lebanon and its staff, it was imperative that we engage with the affected parties. We conducted seven interviews with people from within and outside the City of Lebanon, including staff, department heads, the mayor, and City Councilors representing residents. The insider look provided by these stakeholders gave us invaluable insight to what needed to be included in the revisions and how it would impact the City and its staff. Understanding how the policy manifests for City staff was crucial to tailoring the policy to meet their needs, address their concerns, and have the most positive impact it can.

Project Deliverables

Revised ADM-143

The project produced a revised version of ADM-143, the City of Lebanon’s policy governing the use of artificial intelligence. Building on the existing framework, the revision integrates findings from academic research and feedback from stakeholders to strengthen the policy’s clarity, relevance, and long-term usability. Updates included refined language, the addition of new sections to address emerging needs, and provisions to ensure the policy remains adaptable as AI technologies evolve. The transparency and data privacy sections were significantly expanded to provide clear guidance for AI disclosure and privacy protection. We also added sections on overarching principles, risk review methods, sustainability considerations, and a governance framework to clearly designate responsibility for policy implementation.


AI Guidelines

To supplement the revised ADM-143 policy, we created a set of supplementary guidelines designed to offer more flexible guidance to help employees responsibly implement AI in their daily workflow. The guidelines include a list of sample use cases for AI, responsibilities for AI champions, key skills and competencies for employees, and an AI risk assessment questionnaire for employees to strategically consider the risks and benefits associated with their AI use.


Project Impact and Future Work

Following its review by the TRC and formal approval, the revised ADM-143 policy will be released to all Lebanon city employees. Department AI Champions will be selected by the Chief Innovation and AI Officer, and employee training initiatives will be expanded. This project is expected to meaningfully impact the city of Lebanon and its residents by allowing the city to leverage efficiency gains from AI while preserving fairness, privacy, and human dignity. Because budget is a major concern for residents, AI offers strong potential for expanding city initiatives without creating additional cost. Clear guidance on AI will allow employees to feel more comfortable using this technology and prevent potential errors from arising due to unchecked AI use.

Future work could focus on identifying new ways that AI could benefit the city and developing pilot projects to test new initiatives. One possibility is to explore how AI could help reduce Lebanon’s housing shortage, which was mentioned as a key problem in many stakeholder interviews. AI may be able to streamline permitting and inspector documentation, or identify new locations for planning housing. AI could also help advance the city’s sustainability goals. Other municipalities have deployed AI to help reduce energy use and support infrastructure management, and similar projects might gain strong support from Lebanon’s environmentally conscious population. Lebanon should also consider expanding public outreach related to AI, and partnering with new groups such as business coalitions or nearby academic institutions to expand its resources related to AI and hear more citizen perspectives.

NOTE: The policies above are subject to the City's formal review and approval processes and may be modified before final adoption and publication.

Contributors

Archer Amon (Project Lead)

Georgetown

Annekke van Gelder

George Mason University

Christine Do

UC Davis

Kirin Mohile

Duke University

Maryanne Agyei

Washington University in St. Louis

Navya Sinha

Columbia University

Sana Fathima

Rutgers University

Tatia Jahan

Lafayette College