French publishers face delays and transparency issues in Facebook’s fake-news crackdown

Facebook has partnered with French news organizations to help Facebook verify posts. While publishers are glad Facebook has started the initiative, they say the process lacks efficiency and transparency.

In February, Facebook announced an ongoing partnership between eight French media companies, including Le Monde, Libération and Agence France-Presse, to fact-check and filter news articles that Facebook users flag as false. Once flagged, at least two of the media companies have to agree a story is false. Then, it will be tagged, and Facebook will issue a warning if someone shares it. The process allows Facebook cover to say it is acting as a platform while still showing it is doing something to address the proliferation of propaganda there.

Samuel Laurent, head of Les Décodeurs, Le Monde’s fact-checking team, said this process typically takes between five and seven days. By that point, a flagged post could have been shared thousands of times. “Often some reviews wait several days before an action is taken against the fake news, so it’s probably too late at this point,” he said.

The number of posts flagged by users as potentially false range from about five to 10 a day, said Laurent, adding that 12 people from Le Monde’s newsroom are dedicated to fact-checking. At launch, publishers expressed concerns about how much editorial resource to direct to the project, but resource constraints haven’t been a huge concern with this lag time.

Relevancy has been an issue, according to publishers within the initiative, as many of the stories are hoaxes, rather than highly partisan or opinionated posts that could sway political opinions. “It’s not targeted. We have a list of 100 false claims; about five or six of them will be political,” said Laurent. “The news flagged as fake is often like ‘This dog is a hero, which saved the life of 39 people by biting them’ or ‘Aliens built the Giza pyramids’ rather than ‘real’ — in the sense of having an impact — fake news,” he said. Even though these stories are not particularly relevant, the team spends time debunking them.

Flagging so many hoax stories could be a result of people being on high alert for fake news, which can only be seen as a positive, notes Xavier Grangier, Libération’s head of digital. “Facebook has communicated a lot on the subject, as have other media, so people seem a bit more aware of fake news, and that’s very good.”

France’s fake-news problem hasn’t been as acute as in the U.S. Facebook has faced pressure for its role in spreading false stories ahead of the U.S. presidential election and is also running third-party verification checks there with media organizations. NewsWhip data leading up to the U.S. election found that 72 of the top 200 most-shared stories on Facebook in the U.S. were from fake-news sources. In comparison, 20 out of 200 of France’s most-shared stories on Facebook came from fake news sites, according to NewsWhip data over a two-month period.

In France, Facebook shuttered 30,000 accounts in a fake-new crackdown prior to the French presidential election earlier this month. It also took out newspaper ads in France on how to spot false news. Fake news is potentially more than just a PR problem in Europe, where governments are threatening fines unless action is taken. In Germany, for instance, new legislation threatens Facebook (and other platforms) with fines up to €50 million ($53 million) over disseminating false news items.

Grangier said more transparency in Facebook’s fact-checking initiative is needed: There’s no way to know how many times a post has been mentioned as fake news by other organizations, he said, or any information about whether other fact-checkers are working on a given post. Facebook has not shared any data, either, which could prove the effectiveness of the project.

The initiative also doesn’t necessarily guard against filter bubbles. “It’s quite strange that big fake-news stories that we had previously debunked weren’t flagged on Facebook,” said Laurent. “My hypothesis it that right-wing people are sharing this fake news in their own circles and don’t flag it as false because they think that it’s true.”

Facebook’s team in California is running the initiative. The distance adds a layer of complexity when trying to iterate or make the process more efficient.

Overall, though, publishers are positive about Facebook’s intervention. “This was useful; this experience has helped to keep down the fake news in France,” said Laurent.

“It seems the election has calmed some people,” said Grangier, noting there seem to be fewer false stories on social media, “but I guess those are the sunny days before the storm.”

More in Media

Teads’ M&A rumors are firming up with a deal to merge with Outbrain

The latest installment of ad tech M&A activity is leaving some industry folks surprised.

Publishers’ Privacy Sandbox pauses settle into a deep freeze following reports of poor performance

Publishers aren’t ready to press play yet on dedicated Privacy Sandbox tests.