AI has leveled the field between cybercriminals and nation-state hackers, FBI official says

Assistant Director for FBI's Cyber Division Brett Leatherman (R) speaks with Senior Vice President at Palo Alto Networks Sam Rubin and GovExec Editor-in-Chief Frank Konkel (L) at the Palo Alto Public Sector Ignite conference Oct. 30.

Assistant Director for FBI's Cyber Division Brett Leatherman (R) speaks with Senior Vice President at Palo Alto Networks Sam Rubin and GovExec Editor-in-Chief Frank Konkel (L) at the Palo Alto Public Sector Ignite conference Oct. 30. David DiMolfetta/Staff

The FBI is slower to adopt AI tools due to the sensitive nature of the data that it works with, Brett Leatherman, the FBI Cyber Division head, said.

Criminal hackers, who for years lacked the sophistication and resources of nation-state cyber adversaries, are now on near-equal footing with state-level powers like China and Russia, thanks to advances in artificial intelligence, the head of the FBI’s Cyber Division said Thursday.

“[AI] allows mid-tier actors to really asymmetrically scale in ways that they can’t have impact otherwise, meaning a lot of these cybercriminal groups now have nation-state-type capabilities that they would not otherwise have because they’re using generative AI,” Brett Leatherman said Thursday at the Palo Alto Networks public sector conference in Virginia.

Cybercrime groups are typically financially motivated hacking collectives who seek profit through hacks like ransomware and data theft. Nation-state units are government-backed operators pursuing political, strategic or intelligence-gathering goals. The lines between the two are sometimes blurred.

But nation-state hackers have certainly utilized AI tools as well. Such capabilities have been instrumental for China, a top U.S. cyber adversary, Leatherman added, noting “the [Chinese Communist Party] really is leveraging it to scale their operations right now.”

The FBI has not been as quick to adopt AI in its day-to-day operations because it handles sensitive data that requires stringent protections and oversight to maintain security and legal standards, he said. 

“We’re trying to catch up in many ways, and part of that is because we have very sensitive datasets that we have to make sure we protect because of the authorities that we have,” Leatherman said. “And so keeping those platforms, I guess, honest to what they do with the datasets available to us, is important.”

But there is room to grow. The FBI constantly views data logs and other intelligence collected from legal authorities that can help them track hackers and build computer forensic conclusions. Having AI available to quickly parse those logs would be a benefit, he said, although industry partners are already using their own AI instruments to scan data and report those findings to the FBI.