The FBI is keenly focused on threats emanating from the rapid development of artificial intelligence, particularly from China, top agency officials said Wednesday.
Speaking at the FBI Atlanta Cyber Threat Summit at the Georgia Tech Research Institute Conference Center, FBI Director Christopher Wray and Assistant Director, Cyber Division, Bryan Vorndran warned that while cyberthreats to the U.S. are coming from around the world, China looms largest in both capacity and the ability to weaponize data, in part, with the advent of AI-enabled operations.
Describing China as a threat “unparalleled among foreign adversaries” that has a bigger hacking program than “every other major nation combined,” Wray said that the sheer volume of personal and corporate data stolen by China, mixed with the power of AI, spells big trouble.
“We assess AI will enable threat actors to develop increasingly powerful, sophisticated, customizable, and scalable capabilities — and it won’t take them long to do it,” Wray said. “That goes double for China, which as I mentioned earlier has spent years stealing both our innovation and massive troves of data that’s perfect for training machine learning models.”
China is now “in position to close the cycle — to use the fruits of their widespread hacking to power, with AI, even more powerful hacking efforts.”
Vorndran described a variety of adversarial machine learning attacks, which target the data and systems underlying machine learning models. “The FBI is hyper-focused on these kinds of attacks and specifically on what can — and will — go wrong.”
For now, “highly sophisticated AML attacks, like data poisoning, are still mostly found in research literature and not yet in the wild,” he added. But, he said, “as public and private industry adoption of AI increases, the AML attack surface will correspondingly increase along with the potential physical and economic costs of a successful attack.”
The FBI’s approach to AI and machine learning “is proactive,” Vorndran said, as the agency is “proactively aligning resources to engage with the intelligence community and our private sector partners to better understand the technology and its downstream impacts.”
Concerns about how AI could rapidly increase security issues for the U.S. is something officials in the White House are focused on, as well. Leading AI tech companies recently signed voluntary commitments that they would prioritize safety and security in developing the technology.
Additionally, Anne Neuberger, a top White House official on cybersecurity and emerging technology, recently told CyberScoop that the U.S. government has carried out defensive cybersecurity briefings to leading AI firms regarding their data models, particularly as those firms move away from open source models and seek to close access.