By Josh Keating of The Messenger
The U.S. government is concerned about foreign actors, including China, using new artificial intelligence tools to spread lies and disinformation in the U.S. during the 2024 presidential election, as well as launching cyberattacks aimed at “sowing chaos” in U.S. society, a top Pentagon cyber official said on Friday.
“Certainly [the People’s Republic of China] is one of the actors that we are quite concerned about when it comes to elections defense and foreign malign influence,” Mieke Eoyang, the deputy assistant secretary of defense for cyber policy, said at a reporter’s roundtable hosted by George Washington University. “And I think our concern is that they will see the value in that kind of misinformation/disinformation, and use these tools to get better.”
Eoyang said that generative artificial intelligence capabilities–programs such as ChatGPT which study huge data sets in order to generate plausibly realistic text or images–can be useful as a “means to help people who may have not a particularly wide range of language skill affect a nation where they don’t speak the same language.”
That means that future misinformation campaigns targeting U.S. elections could be cheaper and easier to carry out than Russia’s well-publicized efforts in 2016. They could also be more effective.
Earlier this week, the New York Times reported on a Chinese misinformation campaign which used AI-generated text and images to suggest that the recent wildfires in Maui were caused by a “weather weapon” being tested by the U.S. government. It’s believed to be one of the first large-scale efforts to use artificial intelligence to spread misinformation, and officials worry it could be a preview of what’s in store for the 2024 election.
Traditionally, state-backed online influence campaigns like these have been carried out by large groups like the St. Petersburg-based troll army employed by the late Yevgeny Prigozhin, or the so-called “50-cent army” of online nationalists who push messages on behalf of the Chinese government.
AI could allow governments to generate the same amount of content with far less manpower, and potentially higher believability. A report released by the RAND Corporation this month suggested a scenario in which social networks like Twitter and Facebook would be populated by massive numbers of AI-generated fake accounts that “do not just sound like native U.S. English speakers but use regional variations, such as “Pittsburghese” or Southern American English. They get jokes and U.S. cultural references, and they post pictures of their life: camping with the kids, their dog lying on the living room rug, a birthday party…they also share their political opinions from time to time.”
According to the Times report, U.S. intelligence officials believe Chinese influence campaigns will likely try to undermine support for President Joe Biden and boost the campaign of Donald Trump, but Eoyang wouldn’t comment on Beijing’s intentions.
“I wish I could read Xi’s mind,” she said.
Eoyang also noted that efforts to bolster election cybersecurity have “broad bipartisan support.”
Elections are not the only area of concern for U.S. cybersecurity officials when it comes to China. Earlier this week, the Pentagon released a new cybersecurity strategy, the unclassified version of which warned that both China and Russia “have embraced malicious cyber activity as a means to counter U.S. conventional military power.”
Last May, Microsoft released a report on a state-backed Chinese hacking group nicknamed “Volt Typoon,” which it said has targeted a wide range of government and private organizations and is part of a campaign aimed at disrupting “critical communications infrastructure between the United States and Asia region during future crises.”
On Friday, Eoyang said the Volt Typhoon report and other recent revelations of Chinese activity suggested “a theory of disrupting military mobilization, but also sowing chaos in the United States.”
Others:
