The FBI explains how the Celestial Empire is utilising nefarious machine learning with data that was stolen from the US.
Senior FBI officials expressed particular concern about the threats posed by the rapid development of artificial intelligence, particularly from China, at the FBI Cyber Threats Summit in Atlanta on Wednesday.
According to Christopher Wray, Director of the FBI, and Brian Warndran, Deputy Director of Cyber Security, cyber threats to the United States come from all over the world, but China is the most dangerous adversary in terms of volume and ability to maliciously use data, thanks in part to the introduction of artificial intelligence-based operations.
Ray called China a threat "unrivalled among foreign adversaries." He claims that the country has such a large hacker programme that it can easily bypass "all other major countries combined." Ray also predicted that the sheer volume of personal and corporate data stolen by China, combined with the power of artificial intelligence, would spell disaster.
"We believe that in the future, artificial intelligence will enable attackers to develop increasingly powerful, complex, customizable, and scalable threats - and it will not take them long to do so," Ray said. The FBI director also mentioned that the massive amounts of data stolen by China over the years are ideal for feeding into machine learning models.
According to the FBI, China is now in a position to close the loop, using the fruits of its large-scale hacks to launch even more powerful hacking attacks using artificial intelligence.
Vorndran also described a number of malicious machine learning attacks on various data and systems. "The FBI is overly focused on these types of attacks, specifically on what can and does go wrong."
So far, "high-tech attacks based on malicious machine learning, such as data poisoning, have mostly been found in scientific literature rather than in real-world attacks," Vorndran added. However, he claims that "as the public and private sectors increasingly use artificial intelligence, the attack surface using hostile machine learning will grow in lockstep, as will the potential physical and economic costs of a successful attack."
The FBI's approach to AI and machine learning is "proactive," according to Vorndran, because the agency "proactively aligns resources to engage with the intelligence community and private sector partners to better understand the technology and its implications."
The White House is concerned about how quickly artificial intelligence could increase US security concerns. Leading artificial intelligence technology companies have even signed voluntary pledges to prioritise security and data protection when developing their technologies.