Department of Justice expands use of AI in law enforcement, FBI doubles AI applications in 2025 — FEDagent

Applications of AI


New information about how the Department of Justice (DOJ) is implementing artificial intelligence (AI) tools in law enforcement operations.

Overall, the Department of Justice listed a total of 315 AI entries and 188 active use cases in its 2025 inventory. This is a 31 percent increase from 2024. Law enforcement applications accounted for 62 percent of the use cases deployed, and administrative and IT-related systems accounted for 23 percent.

The Department of Justice noted that its inventory is incomplete and that some information is “withheld due to recognized limitations on information sharing consistent with existing laws, policies, and regulations.”

FBI use case

A closer look at the numbers shows that the Federal Bureau of Investigation (FBI) has doubled its AI applications in the past year.

In 2025, the FBI had a total of 50 AI use cases, 27 of which were classified as law enforcement-related. This is up from 19 overall AI use cases designated to support “law and justice” in 2024.

The FBI is deploying AI for a variety of tasks, including nine tasks designated as high-impact. These include the use of facial recognition and data mapping technologies, biometric matching across large datasets, next-generation location data, and automated vehicle identification.

Low-impact use cases in use include summarization and transcription tools, editing tools, and data extraction tools.

However, the report notes that the FBI has not yet completed the necessary impact assessments on issues such as bias, safety, and performance. The Trump Administration has set an April 2026 deadline for high-impact AI use cases for all federal agencies to meet minimum risk management requirements.

Other Justice Department agencies have also not completed the required assessments, but say work is ongoing.

Privacy advocates raise concerns

The rapid adoption of AI technology and the Justice Department’s reporting methods have raised concerns among privacy advocates.

“There’s some ambiguity in how they chose to describe some of the things in the document itself,” said Patrick Eddington, a senior fellow in homeland security and civil liberties at the Cato Institute. “It’s very difficult to get any kind of final sense of who’s doing the oversight, auditing and stress testing and how good they are.”

The AI ​​inventory is mandated under a 2020 executive order from President Trump’s first term.



Source link