The Snopes and PolitiFact ads show how broad the problem of online misinformation can be, said David Letzler, research scientist at Impact Radius, a digital marketing intelligence firm. “Even websites whose mission is to promote accountability can inadvertently wind up getting used by snake oil salesmen,” he said.
Google declined to explain the specifics of how the fake news ads appeared on the fact-checking sites. The accounts that advertised on Snopes and PolitiFact were terminated from Google’s ad platform after The Times asked about them, according to a person with knowledge of the sites who asked to remain anonymous because the details were confidential.
“As always, when we find deceptive ad practices on our platforms we move swiftly to take action, including suspending the advertiser account if appropriate,” Chi Hea Cho, a Google spokeswoman, said in a statement. “In addition, we give publishers controls so they can block specific types of ads and advertisers.”
When alerted to the ads promoting untrue stories on their sites, Snopes and PolitiFact said there was little they could do. Google’s AdSense, which is used by web publishers to sell display advertising on their sites, works through automated tools. Often, advertisers are unsure where their ads are running — sometimes next to inappropriate or offensive content — and site owners don’t know which ads will appear on their pages.
Vinny Green, Snopes’s co-owner and vice president, said it has tried to filter out misleading advertisements from the 150 million ads it displayed on its site last month. But that only goes so far.
“We have little direct oversight or control over what is being done to filter out fake news ads being served on our site,” he said in an email. He added that the online ad ecosystem is complicit in disseminating and profiting off of misinformation, and that “these ad quality problems are systemic.”
Aaron Sharockman, executive director of PolitiFact, said it was working with Google to remove the “questionable text ads” from its site.
“The revenue those advertisements provide is critical to funding a website like ours, but it’s equally important that we do everything we can to make sure the advertisements appearing on our site are not deceptive or intentionally misleading,” he said.
Google, which sells more online advertising than any other technology firm, has struggled to prevent fraudulent websites from making money off the spread of false stories. Earlier this year, it touted its efforts to crack down on misinformation sites by kicking 340 websites and 200 publishers off its AdSense platform. Most of those publishers had created websites to peddle eye-catching but untrue political stories, and loaded up the pages to get a cut of advertising revenue from Google.
But the websites that advertised on Snopes and PolitiFact took a different approach. Those publishers paid Google to promote their content on legitimate websites to draw traffic for an ad pretending to be a news story, often carrying the banner of a mainstream news publication.
Google has called the process “tabloid cloaking.” It has said these types of scammers use timely topics and ads made to look like new website headlines. Google said it suspended more than 1,300 advertiser accounts in 2016 for tabloid cloaking.
“It’s hard to police and hard to correct but there has to be more measures put in place to fight these actors,” said Marc Goldberg, chief executive of Trust Metrics, an ad safety vendor.
The ad heralding the untrue story about Mrs. Trump’s decision to leave Washington and the White House appeared at the top of PolitiFact last Friday. The display ad led to a fake Vogue news article claiming that the story was also featured on Yahoo, Vanity Fair and Time, among other publications.
On another PolitiFact page, there was another display ad promoting a false story about Mr. Osteen, the leader of Lakewood Church, a Houston-based megachurch, leaving his wife because she had said too much on TV. When clicked, the ad went to a fake Us Weekly page about how Mrs. Osteen was leaving the church to focus on her skin cream company.
The publisher of the fake Us Weekly site also bought a Google search ad, so that a query for the headline of the story “Sad News For Lakewood Church” linked to that same fake story.
In the case of the HGTV stars Joanna and Chip Gaines, the ad about the couple that ran on Snopes last week promoted a report that had been refuted already by the fact-checking site in an article a month earlier.
PolitiFact and Snopes are among the most influential and most popular fact-checking websites. Snopes was created in 1994 to debunk urban legends but has since evolved into a 16-person operation that also assesses political spin. PolitiFact, launched in 2007 as a service of the Tampa Bay Times, won a Pulitzer Prize for its work during the 2008 presidential election and has 14 state chapters in addition to a national operation.
Both PolitiFact and Snopes partnered with Facebook last December, after the social network was criticized for abetting the spread of fake news during the 2016 presidential election. Their experiment — in which viral and popular posts debunked by the fact checkers are marked as “disputed,” and cannot be promoted in News Feeds — expanded and became more aggressive in August.
Google, too, has attempted to dispel misinformation. Jigsaw, its tech incubator, developed a “Share the Facts” tool alongside the Duke University Reporters’ Lab. The widget summarizes fact checks, can be easily shared and, once embedded, carries a fact-check label in Google search and news results.
David Carroll, an associate professor at the Parsons School of Design who has researched advertising technology and fake news, said Google and Facebook ultimately don’t know enough about their customers — advertisers — making it too easy to game the system without serious repercussions.
“For a few bucks, they allow anyone to target anyone else in the world,” Mr. Carroll said. “The whole ecosystem has been infiltrated, and there is insufficient friction to keep bad actors out of the system.”