The Rise of AI Fake News Is Creating a ‘Misinformation Superspreader’

Artificial intelligence is automating the creation of fake news, spurring an explosion of web content mimicking factual articles that instead disseminates false information about elections, wars and natural disasters.

Since May, websites hosting AI-created false articles have increased by more than 1,000 percent, ballooning from 49 sites to more than 600, according to NewsGuard, an organization that tracks misinformation.

Historically, propaganda operations have relied on armies of low-paid workers or highly coordinated intelligence organizations to build sites that appear to be legitimate. But AI is making it easy for nearly anyone — whether they are part of a spy agency or just a teenager in their basement — to create these outlets, producing content that is at times hard to differentiate from real news.

One AI-generated article recounted a made-up story about Benjamin Netanyahu’s psychiatrist, a NewsGuard investigation found, alleging that he had died and left behind a note suggesting the involvement of the Israeli prime minister. The psychiatrist appears to have been fictitious, but the claim was featured on an Iranian TV show, and it was recirculated on media sites in Arabic, English and Indonesian, and spread by users on TikTok, Reddit and Instagram.

Related:Microsoft Is Offering to Help U.S. Politicians Crack Down on Deepfakes

The heightened churn of polarizing and misleading content may make it difficult to know what is true — harming political candidates, military leaders and aid efforts. Misinformation experts said the rapid growth of these sites is particularly worrisome in the run-up to the 2024 elections. 

“Some of these sites are generating hundreds if not thousands of articles a day,” said Jack Brewster, a researcher at NewsGuard who conducted the investigation. “This is why we call it the next great misinformation superspreader.”

Generative artificial intelligence has ushered in an era in which chatbots, image makers and voice cloners can produce content that seems human-made.

Well-dressed AI-generated news anchors are spewing pro-Chinese propaganda, amplified by bot networks sympathetic to Beijing. In Slovakia, politicians up for election found their voices had been cloned to say controversial things they never uttered, days before voters went to the polls. A growing number of websites, with generic names such as iBusiness Day or Ireland Top News, are delivering fake news made to look genuine, in dozens of languages from Arabic to Thai.

Readers can easily be fooled by the websites.

Global Village Space, which published the piece on Netanyahu’s alleged psychiatrist, is flooded with articles on a variety of serious topics. There are pieces detailing U.S. sanctions on Russian weapons suppliers; the oil behemoth Saudi Aramco’s investments in Pakistan; and the United States’ increasingly tenuous relationship with China.

The site also contains essays written by a Middle East think tank expert, a Harvard-educated lawyer and the site’s chief executive, Moeed Pirzada, a television news anchor from Pakistan. (Pirzada did not respond to a request for comment. Two contributors confirmed they have written articles appearing on Global Village Space.)

But sandwiched in with these ordinary stories are AI-generated articles, Brewster said, such as the piece on Netanyahu’s psychiatrist, which was relabeled as “satire” after NewsGuard reached out to the organization during its investigation. NewsGuard says the story appears to have been based on a satirical piece published in June 2010, which made similar claims about an Israeli psychiatrist’s death.

Having real and AI-generated news side-by-side makes deceptive stories more believable. “You have people that simply are not media literate enough to know that this is false,” said Jeffrey Blevins, a misinformation expert and journalism professor at the University of Cincinnati. “It’s misleading.”

Websites similar to Global Village Space may proliferate during the 2024 election, becoming an efficient way to distribute misinformation, media and AI experts said.

The sites work in two ways, Brewster said. Some stories are created manually, with people asking chatbots for articles that amplify a certain political narrative and posting the result to a website. The process can also be automatic, with web scrapers searching for articles that contain certain keywords, and feeding those stories into a large language model that rewrites them to sound unique and evade plagiarism allegations. The result is automatically posted online.

NewsGuard locates AI-generated sites by scanning for error messages or other language that “indicates that the content was produced by AI tools without adequate editing,” the organization says.

The motivations for creating these sites vary. Some are intended to sway political beliefs or wreak havoc. Other sites churn out polarizing content to draw clicks and capture ad revenue, Brewster said. But the ability to turbocharge fake content is a significant security risk, he added.

Technology has long fueled misinformation. In the lead-up to the 2020 U.S. election, Eastern European troll farms — professional groups that promote propaganda — built large audiences on Facebook disseminating provocative content on Black and Christian group pages, reaching 140 million users per month.

Pink-slime journalism sites, named after the meat byproduct, often crop up in small towns where local news outlets have disappeared, generating articles that benefit the financiers that fund the operation, according to the media watchdog Poynter.

But Blevins said those techniques are more resource-intensive compared with artificial intelligence. “The danger is the scope and scale with AI … especially when paired with more sophisticated algorithms,” he said. “It’s an information war on a scale we haven’t seen before.”

It’s not clear whether intelligence agencies are using AI-generated news for foreign influence campaigns, but it is a major concern. “I would not be shocked at all that this is used — definitely next year with the elections,” Brewster said. “It’s hard not to see some politician setting up one of these sites to generate fluff content about them and misinformation about their opponent.”

Blevins said people should watch for clues in articles, “red flags” such as “really odd grammar” or errors in sentence construction. But the most effective tool is to increase media literacy among average readers.

“Make people aware that there are these kinds of sites that are out there. This is the kind of harm they can cause,” he said. “But also recognize that not all sources are equally credible. Just because something claims to be a news site doesn’t mean that they actually have a journalist … producing content.”

Regulation, he added, is largely nonexistent. It may be difficult for governments to clamp down on fake news content, for fear of running afoul of free-speech protections. That leaves it to social media companies, which haven’t done a good job so far.

It’s infeasible to deal quickly with the sheer number of such sites. “It’s a lot like playing whack-a-mole,” Blevins said.

“You spot one [site], you shut it down, and there’s another one created someplace else,” he added. “You’re never going to fully catch up with it.”