When I was first sworn in as ninth director of the , one of my top priorities was to modernize the bureau with new, cutting-edge technology that would allow us to better serve and protect the American people. When I arrived, the FBI was running on archaic patchwork systems without AI, effectively putting a 2025 car battery into a vehicle from 1985. Our infrastructure was a Commodore 64 when it needed to be a supercomputer. No more Band-Aids on gunshot wounds. Wholesale change was necessary.
In the past year, Im incredibly proud of the progress weve made under President Trumps leadership. We have rebuilt and revamped the FBIs infrastructure across the enterprise, helping the bureau achieve record-breaking results in crushing violent crime and defending the homeland, while providing historic transparency.
Artificial intelligence is a huge part of that overhaul. When then-Deputy Director and I arrived here at headquarters, AI had almost zero role at the FBI. That had to change, so we got to work. We immediately led the way by setting up an AI working group to evaluate how we could accelerate modernization, getting input from field leaders on the ground in your communities. We appointed a chief AI officer and established an AI Review Board to streamline our efforts. We created an AI Champions Program to identify advocates across the bureau. Maybe most importantly, we created direct partnerships with private-sector industry leaders to rebuild our infrastructure and bring in AI on a broad scale.
AI is central to what we do. It is helping us identify victims of predators, and more. Last year alone, this FBI identified and located 6,300 missing kids, a 30% increase, and arrested 2,000 abusers, a 20% increase â largely thanks to these improvements. In a recent FBI Richmond case, the FBIs Child Exploitation Operational Unit used facial recognition tools to save 8- and 12-year-old children from a would-be abuser, who will now spend 50 years in prison.
This FBI now uses new AI tools to generate call transcriptions, provide concise synopses and even help correlate contacts with other received complaints. When someone calls into NTOC â the National Threat Operations Center, our 911 center â AI tools generate a transcript of the call, draft an effective summary of the threat and immediately scan our database for comparisons to other open threat lines. Every tip also receives a lead value to surface the highest threat-related calls for Threat Intake Examiners. This specific threat intake process helped the FBI quickly act and stop an attacker plotting a at a North Carolina preschool.
Fingerprint matching is one of the most common methods the FBI uses to identify individuals. Some adversaries try to manipulate their prints to obscure their identities by burning, cutting or biting their fingertips to remove ridge detail and make it difficult or impossible to make a match. Because fingerprint matching is an automated process, altered fingerprints were being missed. CJIS â the Criminal Justice Information Services Division, our data hub â integrated an AI-enabled, real-time altered fingerprint detection capability. In 2025, this new solution detected 34 altered fingerprint identities, ultimately leading to the positive identification and arrest of wanted persons, drug traffickers and fraudsters.
Were also using AI to facilitate rapid translations of large volumes of text, audio and video, and to triage terabytes of data. In the aftermath of the Oct. 7, 2023, attacks on Israel, the FBI had more than 75 terabytes of data to review, with more information coming in every day, including more than 75 search warrant returns. For perspective, a single search warrant return can contain 180,000 messages. It would take six or seven analysts working seven days a week for four or five weeks to review a single search warrant return. The FBI routinely has thousands of audio files and texts to review over the course of a s
