FBI Intel Strategy Includes Using AI to Process Terabytes of Data

The new FBI intelligence program strategy will use technology to process information with speed and accuracy, FBI Director Christopher Wray said while announcing the plan. Speaking at the Intelligence and National Security Summit on Feb. 29 in McLean, Virginia, Wray emphasized the importance of artificial intelligence and other emerging tech to the five-year strategy.

“Just one investigation of a cyber intrusion could bring in terabytes of data,” said Wray. “And we have to quickly distill that information so we can find the needle in an ever-growing haystack.”

The FBI can use evolving technologies like AI and machine learning to sift through data more efficiently, Wray said. However, relying on intelligent automation to distill information can raise ethics and accuracy concerns. Officials said that the use of AI will continue to rely on human oversight.

“We’ve been very keen on ensuring that we can use AI and analytics for things that have a human still on either end,” said Cynthia Kaiser, deputy assistant director for FBI’s Cyber Division at the CyberScape Summit on March 8 in Reston, Virginia. “It’s going to make us much more efficient and still allow us to put those protections in place and ensure we’re doing right by the American people and we’re not jumping to conclusions because we automated everything. We have to make sure there’s some validations in place because it’s important for us to be right”

Having a workforce that can analyze information gleaned from AI is critical to the FBI’s mission and its intelligence strategy.

“A deluge of data is overwhelming human capacity to process, exploit, assess, share and act on it and in a timely way,” Wray said. “[The Office of the Director of National Intelligence] was forward thinking in identifying the need to augment our greatest resource our human talent, with tools and technology to reduce the risk of missing important threat signals amid all the noise.

While innovative technologies can enable the U.S. to do high-quality work more quickly, they can also endanger the country when they fall into the hands of adversaries.

According to Kaiser, the FBI is focusing on some major risks associated with AI.

“[A risk of AI advancement] is a non-rule of law nation like the People’s Republic of China being able to actually steal our AI advances,” said Kaiser. “We have great technology companies within the U.S. doing a lot of amazing development, and we want to keep it here. We want to make sure that the people who develop that technology are able to continue to pursue that. We’re focused on ensuring that we protect our American ingenuity from adversaries who want to profit off of it and steal it.”

Adversaries beginning to experiment with AI is another primary concern. By spending more time with new technologies, they have learned to work more efficiently, improve the believability of frauds and hide in networks longer, Kaiser said.

“AI probably has more cybersecurity benefits than drawbacks from adversary use, but that could change,” said Kaiser. “That’s why the FBI is being so forward-leaning, ensuring that we’re partnering with emerging technology companies, AI companies, as well as partnering just across, and continuing to do the work we’ve always done in the cyberspace … so that they don’t get better.”

Using technology to counter adversarial AI is not just the future, Wray said, but the bureau is already using AI to stay ahead of evolving threats.

“We are doing plenty of that, whether it’s improving the speed with which we can translate things that are in foreign languages or getting through all that deluge of data,” Wray said. “AI is very useful in helping to identify intrusions and adversarial use of AI.”