FBI is using much more AI than what it previously disclosed

After more than doubling its artificial intelligence use cases from 2024 to 2025, it appears the FBI is on track to sustain or increase its pace again in 2026. 

The bureau has a current use case tally totalling 100-plus, according to FBI Chief AI Officer Katie Noyes. In its AI inventory posted in January, the Department of Justice unit identified just 50 use cases. 

“We’ve got, by the way, 139 approved use cases — all different manners of data, all different manners of processes and certainly all manner of risk,” Noyes, who took over the top AI role last summer, said during the 2026 DODIIS Worldwide Conference this week. “There are some of those that are going to be considered high-impact.”

While Noyes did not provide additional information about the focus of the new use cases, the bureau has historically directed the majority of its efforts toward AI that aids law enforcement. 

“Additional case use [sic] approvals will be reflected in the next annual inventory,” an FBI spokesperson told FedScoop via email Thursday. 

Risk management

DOJ keeps a pretty tight lid on the inner workings of its AI plans. Sources previously told FedScoop that the agency’s level of transparency regarding its use cases falls behind those that are also notoriously taciturn, like the Department of Homeland Security. 

Of the FBI’s nine high-impact, deployed AI use cases that were disclosed in January, the bureau had not completed any of its risk management requirements. Federal agencies had a deadline of April 3 to get high-impact use cases up to speed with provisions that were outlined by the Office of Management and Budget. 

DOJ was one of the agencies that missed the deadline — and still hasn’t posted an updated inventory, unlike some of the other laggards. 

Internally, the bureau has shifted how it is thinking about AI-related risk, according to Noyes. 

“We’ve started bringing more of a risk framework and then breaking it down by application,” she said. “We’re getting a lot of really good outputs where we can take a little more risk and we have a little more trust and [then] those where we have to go back to the drawing board because it’s just not ready.”

Each use case goes through an AI review board that handles testing in a variety of environments, per Noyes. 

Future-proofing

The FBI has prioritized building use cases that are model agnostic as a way to keep up with the ever-evolving vendor landscape. 

“We can plug and play,” Noyes said. “And we’ve got the infrastructure.” 

The cost of the technology has posed challenges, pushing the bureau to lean on cloud providers as a go-between rather than heading straight to the source. 

“If I’m being candid, the price point for us right now in the FBI, it’s a little steep,” Noyes said. “We, like many others, have multiple cloud providers and that’s one way you can get the tokens and the processing … and all the select models.”

The bureau is also prioritizing on-premise AI workflows where applicable, AI services through partnerships and hybrid deployment opportunities. 

“Hopefully we’ve done enough to really future-proof,” Noyes said. “And we’re ready for the next thing.”