New ‘Justice AI Tracker’ watches how police, courts are using AI

A research and policy lab at Georgetown University last week introduced an interactive web tool designed to bring transparency to how artificial intelligence is being used in the criminal justice system.

The Evidence for Justice Lab, at the Washington, D.C., university’s McCourt School of Public Policy, published the Justice AI Tracker, or JAI-T, documenting where and how AI technologies are being piloted, tested or implemented in law enforcement, courts and corrections. The tracker aggregates publicly available information on AI use, offering policymakers, researchers, technologists and community advocates a clearer picture of how automated systems are shaping criminal justice outcomes.

The first release focuses on the 100 largest U.S. cities. The next update is expected in six months.

Andrea M. Headley, faculty director at the Justice Lab, said the tracker is intended as both a research tool and a platform for accountability on debates about technology in policing and courts.

“There’s not this reliable, up-to-date way to see where tools are being used,” Headley said in an interview. “I hope that that will change with what kind of AI inventories at the federal government level. I’m hoping that more cities will kind of take up the mantle to be proactive in publishing these things on their own, but in the meantime, we see this as trying to fill that gap.”

Vinuri Dissanayake, deputy director at the Justice Lab, said the project originated from a separate national study to understand how artificial intelligence is perceived in various communities and how the emerging technology is influencing how residents experience interactions with public safety.

“What’s been really fascinating — we’re still in the midst of this study — is the lack of awareness of what people see as being part of their community,” Dissanayake said. “How do we ensure that the communities that are interacting with this tool are A, aware of these tools and B, understand that there is recourse when they feel like it’s not being used in the right way?”

Dissanayake said researchers found five key AI use cases: facial recognition, gun detection, automatic license plate readers, body camera footage review and non-emergency dispatch assistance.

“We’re also seeing AI agents that are being used in different scenarios, from the courtroom to public-facing services, that are occurring in law enforcement,” Dissanayake said, adding to the list of AI use cases.

Dissanayake said future iterations of the JAI-T database will include vendor information and qualitative insights from residents. She said the lab also plans to release a report on the tracker in two months, and that another, more detailed report on AI usage trends in 100 cities is in progress.

“There are obviously some examples of vendors that have been working in this space for a while, but there are a lot of startups that are also working with police departments and developing tools,” she said.