Villa Park trustees hear staff briefing on draft AI policy, no formal action taken

Village Manager Matt Harline led a Committee of the Whole discussion April 14 on proposed uses of artificial intelligence and a draft AI policy for the Village of Villa Park, emphasizing risks to confidentiality, algorithmic bias and the need for human oversight.

Harline said the goal of the session was to outline potential uses and risks and to solicit questions; he noted a draft policy had been distributed to trustees in advance. "Our discussion tonight is about artificial intelligence uses and policy recommendations for the Village Of Villa Park and staff will lead a discussion about the risk associated with the use of artificial intelligence in the public sector," Harline said.

The presentation described common municipal AI uses already in place or under consideration, including road-assessment and traffic-counting software, license-plate readers, Axon redaction tools for body-camera footage, and a civic website virtual assistant. Harline gave examples of generative AI such as ChatGPT and said tools vary from simple autocorrect to systems that incorporate user inputs into their training databases.

Harline and trustees discussed several risks the draft policy aims to address: loss of confidential data if open-source tools ingest municipal information; algorithmic bias that could affect employment screening; lack of transparency in proprietary algorithms; workforce impacts as automation replaces manual tasks; and physical-safety risks where robotics or autonomous systems are involved. He cited a state-level effort and related legislation, noting in the meeting that an Illinois task force report contributed to a House bill referenced in the presentation as taking effect Jan. 1 and restricting AI uses that produce discriminatory outcomes in employment screening.

Harline outlined recommended policy provisions the draft contains or will contain after revision: protect confidentiality and avoid submitting personally identifying or proprietary information to third-party AI services; validate accuracy and fact-check AI outputs; maintain human verification for final decisions; prioritize data quality over quantity; and treat AI-produced content as the user's responsibility when used in official work.

Trustees asked clarifying questions. Trustee Kumar said he uses AI tools for programming and productivity and supported cautious adoption, but raised hiring and discrimination concerns: "I'm not % against AI ... security certainly is gonna be a problem," Kumar said, asking whether Village hiring processes and vendors already use AI screening tools. Harline replied that vendors used for HR or recruitment might incorporate AI and that the Village should verify vendor practices as part of policy implementation.

A member of the public raised concerns about employees and volunteers using personal devices. The commenter asked whether personal computers or smartphones used for Village work could expose municipal systems or data to AI services and third-party databases, describing a scenario in which information typed into a personal account could be absorbed into an AI provider's dataset. Harline explained the Village separates guest WiFi from internal systems, uses VPNs and cloud protections for staff access, and employs blocked-site lists and firewall monitoring, but acknowledged that improper use of public AI tools from a Village workstation or cloud account could pose risks.

Harline noted some operational details mentioned in the presentation: the police department has discontinued use of Grammarly for official reports because text entered into that service can be incorporated into its database; the Village's cyber insurance does not currently include a specific AI exclusion or inclusion and AI-related exposures would fall under existing cyber coverage; and the Village's insurance pool has AI under review for the upcoming renewal on 2025-11-01. He said staff plans to bring a draft policy forward as a resolution after further internal review and consultation with other municipalities and legal/IT advisers.

Votes at a glance

The Committee held no vote on the AI policy. The only formal motion recorded before adjournment was procedural: Trustee Cordova moved to adjourn the Committee of the Whole; Trustee Kumar seconded; the motion passed on a voice vote.

The meeting then closed and the regular Board meeting was scheduled to begin at 7 p.m.