Savannah Alderman Nick Palumbo has proposed a four-pillar framework to govern how the city uses artificial intelligence and what it demands from technology vendors. The draft, shared with council in early August, arrives as city leaders acknowledge that internal policies have not kept pace with the speed of AI development.
Mayor Van Johnson said at a July workshop that the city is "behind" in building an advanced AI governance structure. City Manager Jay Melder confirmed Savannah already has policies for existing tools, such as its Microsoft Copilot license, but lacks a broader framework for how AI will factor into operational and funding decisions. "I think those are the bigger points related to AI governance," Melder said.
A framework built on transparency and human control
Palumbo's proposal rests on four principles: the city must publish what AI systems it uses, humans must sign off on consequential decisions, Savannah's data must remain under local control, and the city must rigorously question technology providers before signing contracts. The alderman said the intent is to protect residents' data while making government use of AI visible to the public.
Specific items include publishing a list of deployed AI systems, their functions, and which departments use them. The proposal also requires the city to vet AI tools for privacy risk, bias, and whether resident data is used to train models. "Ultimately, it's the council that should decide what is shared and what is not in my mind," Palumbo said. "This has to be a public debate, an agenda item, just like a procurement contract."
National context and local urgency
Mayor Johnson served on a National League of Cities AI for Government advisory committee in 2024. The resulting NLC report and toolkit highlights both the promise and pitfalls of municipal AI use, from predictive maintenance on water pipes in Tempe, Arizona, to faster pothole detection in Memphis, Tennessee. It also points to existing policies in cities like San Jose, California, which publishes its generative AI guidelines, and Boston, where human fact-checking of AI-generated content is standard practice.
The report's co-chairs, Johnson and Tucson Councilmember Nikki Lee, wrote that there is "no one-size-fits-all approach to AI adoption in cities." Johnson said in July he wants Savannah's framework completed "soon." Palumbo said he hopes his draft will "get the ball rolling" and expects a future council workshop on the topic.
Why this matters for government, IT, and development professionals
For teams inside city government or vendors selling to municipalities, Palumbo's proposal signals the specific compliance requirements likely to spread across local agencies. Mandatory AI system registries, privacy risk audits, and human-in-the-loop approval gates are moving from policy papers to procurement checklists. Professionals who design or sell AI tools to public-sector clients should prepare to answer detailed questions about training data provenance and bias testing before contracts are signed. The AI Learning Path for Policy Makers addresses these exact governance challenges that cities are now confronting.
Your membership also unlocks:
