Advertisement

FBI expands AI-powered biometric, facial recognition capabilities, inventory shows

In the past year, the Department of Justice division has added 31 AI new use cases with nearly half enabling law enforcement activities.
Listen to this article
0:00
Learn more. This feature uses an automated voice, which may result in occasional errors in pronunciation, tone, or sentiment.
A view of the J. Edgar Hoover Building, which served as the headquarters building for the Federal Bureau of Investigation since 1975, in Washington, DC, on Dec. 29, 2025. FBI Director Kash Patel announced on December 26 that the FBI has finalized a plan to permanently close the building and move the FBI workforce to a new facility. (Photo by Brendan SMIALOWSKI / AFP via Getty Images)

The FBI more than doubled its AI use cases in the past year, according to the Department of Justice’s 2025 inventory posted Friday. Many of those uses are enabling law enforcement activities, from biometric and facial recognition capabilities to data synthesis and triage. 

In 2024, the FBI had 19 AI use cases overall with 15 designated as helping to aid “law and justice.” Last year, the DOJ division accumulated 50 AI use cases total, and 27 of those were categorized as law enforcement. 

“We don’t really have insight into the quality of AI that they’re using,” Patrick Eddington, senior fellow in homeland security and civil liberties at the Cato Institute, told FedScoop. 

None of the high-impact use cases, which are all part of law enforcement processes and have been deployed, have completed the required risk management steps as they quickly approach an April deadline. All except one of these use cases are powered by unnamed vendor-built systems. 

Advertisement

“There’s a level of ambiguity in how they’ve chosen to describe some of the things in the document itself,” Eddington said. “It’s extremely difficult to get any kind of definitive sense in terms of how good it is, who’s conducting the oversight or auditing or stress testing.”

“From what I’ve seen so far, it gives me a sense of concern, unease [and] alarm,” Eddington added. 

The FBI started about five new AI projects meant to generate investigative leads using suggested facial matches and other data as it worked to expand biometric capabilities. Four of those new systems are already actively being deployed in operations. None have additional details about managing potential risks.

“Unfortunately, DOJ, for a second year in a row, seems to have repeated many of the shortcomings we saw last year — primarily that’s their complete lack of any reporting about their implementation of risk management practices,” said Quinn Anex-Ries, a senior policy analyst focused on equity and civic tech at the Center for Democracy and Technology. 

The DOJ even falls behind the Department of Homeland Security’s AI inventory on transparency, a traditionally tight-lipped organization, according to Anex-Ries. “It’s concerning that this is happening multiple years in a row,” Anex-Ries added. 

Advertisement

The three stakeholders FedScoop spoke with all highlighted areas of concern around the comprehensiveness and accuracy of the FBI’s AI inventory. The DOJ did not respond to a request for comment prior to publication, and the FBI did not provide additional information before publication. 

Stakeholders also raised concerns over how tools and use cases were labeled or bunched together. 

“They are quite vague in what they describe and in what the outputs are … it’s like ‘search tool, chatbot, data outlier detection,’” said Valerie Wirtschafter, a fellow in the Brookings Institution’s foreign policy, artificial intelligence and emerging technology initiative. “That stood out to me quite a bit.”

The three clearly labeled instances of facial recognition technology, for example, all have 2025 deployment dates, despite reports that the DOJ division has been using the technology for years. 

“Many advocates and watchdogs have been raising questions over the last several years about the thoroughness of inventories,” Anex-Ries said. “The fact that we’re seeing some tools show up for the first time when we know they’ve been used longer than that definitely should keep us on alert.”  

Advertisement

The AI inventory process dates back to a 2020 executive order during the first Trump administration that was later enshrined into federal statute. The public reporting has had expected delays this year due to the federal government shutdown last year, but most agencies have begun to post their 2025 AI inventories this week. 

While hiccups and gaps remain, the inventories do offer a peek behind the curtain on how taciturn federal agencies are moving forward on AI adoption — for better or worse. 

“When you start to take that extremely broad, totally unregulated investigative authority, and you combine that with the use of these AI tools and all of these other programs and databases that are available, it gives them an amazing ability to put together very complete, or relatively complete, profiles on just about anybody in this country,” Eddington said.

Latest Podcasts