Facebook CEO Mark Zuckerberg has long insisted that the company is politically neutral, but the same apparently can’t be said for all of its content moderators.
The social-media giant said it would review its training and oversight after a Project Veritas sting released Tuesday showed moderators on hidden-camera footage claiming that they delete pro-Trump and conservative posts; that the company harbors an anti-Trump bias, and that Facebook shadow-bans content, despite its claims to the contrary.
In an email, a Facebook spokesperson said, “Our reviewers work hard to keep our platform safe for billions of users every day.”
“The reviewers’ decisions are regularly evaluated to ensure our policies are applied fairly and accurately,” said the spokesperson. “We do not promote political viewpoints of any stripe and the comments included in this video are not consistent with our policies. We will be reviewing our training and oversight to make sure these decisions are made consistent with our policies.”
The Project Veritas video featured an interview with former content moderator Zach McElroy of Cognizant, a third-party firm used by Facebook to police the platform, who said that about three-quarters of the posts flagged for “civic harassment” tilted to the right.
“I saw a stark contrast between Republican versus Democrats in that queue,” said Mr. McElroy. “I saw upwards of 75% to 80% of the posts in that queue were from Republican pages, politicians, journalists, and pages that supported the president or supported conservatives.”
He said the content moderators, who lean heavily to the left, “are essentially in charge of what gets said and what gets stifled.”
“And in speaking with a lot of them, I found that they are not at all shy to exercise their political will in deleting or leaving up content, whether or not they’re allowed to or whether or not they’ll get penalized for it, especially when they’ve got nothing to lose, that is to say, they’re getting laid off,” Mr. McElroy said.
In October, Cognizant announced that it would scale back its content-moderation services in 2020 for Facebook, cutting about 6,000 jobs, following reports in The Verge and the Tampa Bay Times about the stressful working conditions experienced by staffers who monitor graphic and violent posts.
The video featured Project Veritas investigators speaking in hidden-camera footage to about a dozen content moderators, most of whom were identified by name. The Washington Times has not confirmed their identities independently.
One man described as a moderator in Austin, Texas, called Facebook “a very progressive company who’s very anti-MAGA,” referring to President Trump’s Make America Great Again campaign slogan.
While Facebook has a hands-off policy on Mr. Trump’s posts, he said, “the fortunate thing is even if he does say something, if it gets repeated, we can at least get the average Joe.”
Another content moderator said, “I think they’re very biased about who they’re going to protect,” adding that he believes that Facebook shadow-bans, saying, “it’s clear that people’s content don’t come up because it’s been de-filtered off the queue.”
“They’re doing something, man. They’re just trying to pretend that they’re not,” he said.
Several moderators expressed anti-Trump sentiments. “Time to get the Cheetoh out of office,” said one staffer.
Another man said that “like half the time when I delete people for pro-Trump sh*t, I’m like, you should be on a watch list, dude,” while a woman said that on her last day with the company, “If someone is wearing a MAGA hat, I am going to delete them for terrorism and just going to like go crazy.”
Mr. McElroy pointed to the disparate handling of two politically charged graphics: He said a drawing showing Mr. Trump bleeding from the mouth with a knife to his neck was allowed to remain on the platform, while a cartoon of Elmer Fudd “shooting” Democrat Beto O’Rourke was removed.
“There’s no difference between the two,” Mr. McElroy said. “There’s no logical reason why the determined reaction to those two posts would be different. They should be the same.”
He said Facebook also made an exception to its policy against accusations of terrorism with a comment by CNN’s Don Lemon, who said in 2018 that white men are “the biggest terrorist threat in this country,” ruling that the remark was newsworthy.
Donald Trump Jr. tweeted that the Project Veritas investigation raised concerns about 2020 election meddling by social-media platforms, saying, “Enough is enough we can’t let them interfere with elections any longer.”
At an April 2018 hearing, Mr. Zuckerberg told Senate Republicans he takes political bias allegations seriously and has tried to make sure “that we do not have any bias in the work that we do.”
Mr. McElroy acknowledged that he had signed a non-disclosure agreement covering his work on Facebook content, but added, “I think the truth is more powerful than any NDA.”
Known for its hidden-camera investigations, Project Veritas has been accused of deceptively editing its undercover videos, which PV president James O’Keefe has adamantly denied.
“The actions of the content moderators may not be conclusive evidence of structural bias at Facebook, but it certainly speaks to the anti-conservative culture that McElroy says is pervasive,” said Mr. O’Keefe on the video.