St. Petersburg officials are openinga public conversation about artificial intelligencejust days after the city quietly adopted its first policy governing how employees can use the technology.
The City Council unanimously voted Thursday to refer a proposed citywide AI strategy to its Public Services and Infrastructure Committee, while the Welch administration launches a public comment period and open house on the city’s future use of AI.
Council Member Mike Harting requested the referral to the Public Services and Infrastructure Committee, asking members to examine how the city should use artificial intelligence, what guardrails should accompany it and what it could cost.
The council’s discussion came three days after the city adopted its first administrative policy governing generative AI, according to an internal document obtained by theCatalyst. Effective Aug. 3, Administrative Policy 071200 authorizes employees to use approved tools, including Microsoft Copilot, ChatGPT and Gemini, while establishing rules for their use.
The policy prohibits employees from entering confidential information, personally identifiable information, criminal justice records and other protected data into AI systems without written approval.
It also requires employees to verify AI-generated content for accuracy and intellectual property concerns, preserve qualifying AI-generated records and use only city-approved AI platforms.
Employee guidance accompanying the policy says AI may be used for drafting routine communications, summarizing public information, brainstorming ideas and conducting research, but cannot replace employee judgment or be used to make legal, disciplinary, procurement or public safety decisions.
Public-facing content must also be reviewed by a city employee before release.
Not everyone on the council believes the City’s safeguards go far enough.
“I really appreciate the new business item,” Council Member Richie Floyd said. “I think we’ll probably have disagreements on this as it goes forward.”
Floyd questioned whether the city is already paying for enterprise AI licenses, warning that costs can escalate quickly. He also criticized the newly adopted policy, saying he had reviewed it and believed it was “far too permissive.”
“I think we need to temper it back already,” he said.
Floyd questioned whether relying on employees to review AI-generated work was sufficient.
“The policy says everyone must review it, but I don’t think those words ‘everyone must review it’ are strong enough,” Floyd said. “Technical data through generative AI will have hallucinations and whatnot that will be difficult to catch.”
He argued that reviewing AI-generated technical information can sometimes defeat the technology’s intended efficiencies.
“Often times it takes more time to catch those mistakes than it does to just create a correct representation of data in the first place,” Floyd said.
Council Member Copley Gerdes said he has been working with staff and an outside consultant on how AI could improve the city’s permitting process.
“I think one of the places that it could be very helpful is with public records,” Gerdes said, pointing to the city’s ongoing challenges responding to records requests.
Gerdes said he expects to bring a presentation to the committee this fall.
The council’s discussion comes as the Welch administration launches a 30-day public comment period on artificial intelligence and prepares to hostan Aug. 25 open house at the President Barack Obama Main Library. According to the city, residents’ feedback will help shape how AI is used in municipal government.
Mayor Ken Welch has publicly encouraged exploring the technology. Followinga June meeting with technology and community leaders, Welch said he looked forward to continued conversations about using artificial intelligence “to strengthen our City teams and services.”
