Facebook is taking more steps to punish page owners outside of the US that target Americans with financially motivated hoaxes and misinformation.
The company told BuzzFeed News it has deployed machine learning that it says can automatically identify and reduce the reach of pages administered by people who target others elsewhere in the world with hoaxes in order to make money. This approach is squarely aimed at bad actors such as the now-famous Macedonians who run pages about US politics that targeted Americans with fake news during and after the 2016 US presidential election.
On Thursday, Facebook will announce this and other new measures to combat misinformation on its platform in a blog post and at the Global Fact fact-checking conference underway in Rome.
Facebook works with third-party fact-checkers in 14 countries who review content spreading on the platform that’s been flagged as potentially false. Once a checker rates a link, image, or video as false, Facebook reduces its reach and surfaces the resulting fact check as related content in the News Feed.
As part of its announcement today, the company is also expanding a program that enables its fact-checking partners to debunk videos and images containing misinformation, and it is deploying technology to automatically identify and demote duplicate versions of false stories.
“Copycat hoaxes have been an increasing trend in 2017 and also into 2018,” Tessa Lyons, a product manager in charge of News Feed integrity initiatives, told BuzzFeed News in a phone interview. “Using machine learning we’re able to identify and demote duplicates of articles that were rated false by fact checkers.”
The new technology’s focus on foreign-operated Facebook pages and duplicate hoaxes could deal a significant blow to spammers based in countries such as Macedonia, Kosovo, Pakistan, and elsewhere who target English-language Facebook users. BuzzFeed News has documented how these countries have become home to clusters of pages that spread clickbait and misinformation about politics and Native Americans, among other topics. They also frequently plagiarize hoaxes and other content for their associated websites.
Lyons said Facebook’s new machine learning technology will help it identify pages that are “likely” to spread financially motivated hoaxes based on past behavior and other signals.
“These pages often copy and paste content [from other sources], and another signal is that the website themselves are covered in low-quality ads,” she said. “We also see a common pattern in that page admins based in one country are targeting people in other countries. These admins often have suspicious accounts that are not fake but are identified in our system as having suspicious activity.”
Lyons acknowledged that false positives may occur, which could result in a page that isn’t spreading financially motivated hoaxes having its reach reduced.
“We feel pretty good about the fact that these pages [being affected] are the ones that we are intending to target,” she said.
She also said a page that shared hoaxes can earn back its reach if it stops sharing false content.
“There is that ability to kind of rehabilitate [your page],” she said.