The Wizard & I: AI Use in Public Safety Communications and Law Enforcement

August 3, 2026 | 2:00 - 3:00 p.m.

Don Zoufal, attorney, former first deputy director, Chicago Office of Emergency Management and Communications

Jonathan Lewin, deputy chief of support services at the University of Chicago Police Department

Keith Kelley, deputy chief, Athens-Clarke County Police Department, Athens, Georgia

The AI revolution is ongoing in public safety communications. Two members of law enforcement and an attorney discussed how agencies can harness the revolution and maintain appropriate controls as they do.

Zoufal said agencies have relied on AI-driven tools for 40 to 50 years, but generative models mark a shift from systems that describe what already happened to systems that generate original analysis and content. "Traditional AI is the chef that follows a recipe," Zoufal said. "Generative AI is the chef creating new dishes."

Lewin emphasized the pace of adoption for the new technology. He said ChatGPT reached 100 million users in two months and passed 1 billion users by late July. It took 75 years for the telephone to cross the 100 million-user threshold and seven years for the internet to reach similar adoption. He said that this speed, combined with free public access, means officers and telecommunicators are likely already using these tools, sometimes without departmental authorization.

Keith Kelly described how Athens-Clarke County has deployed AI in a real-time crime center, including Flock camera analytics feeding roughly 2,000 camera feeds, patrol vehicles running software that adjusts emergency lighting to de-escalate scenes, drone-as-first-responder operations, and body-camera report-drafting software.

Lewin demonstrated "vibe coding," using conversational prompts to build working dashboards and crime-mapping tools without writing code himself. He cautioned that the process requires repeated verification since models can fabricate sources or data.

When developing agency policies and guidance, Zoufal pointed attendees to publications by the National Institute of Standards and Technology, the General Accounting Office, and the Office of Management and Budget for procurement and policy language. He noted that state legislation on AI is expanding faster than federal rules, and federal law doesn't generally control state and local agencies' AI use. He said departments should start by implementing an AI use policy restricting use to authorized tools in proper environments.

The session closed with a video that Lewin created on Grok that generated anthropomorphized AI models. Gemini moderated a simulated debate between ChatGPT and Grok on AI-assisted policing risks. Using answers that each model generated for the same questions, the debate illustrated the different approaches that the models use regarding questions of transparency and speed in AI-assisted policing.

By Rick Goldstein