Advertisement

Spanish-language misinformation about vaccines is evading Facebook’s moderators

Spanish language posts boasting false claims that the covid-19 vaccine causes magnetism are avoiding fact-checkers at a higher rate than English language posts.
In this photo illustration, a smart phone screen displays the logo of Facebook on a Facebook website background, on April 7, 2021, in Arlington, Virginia. (Photo by OLIVIER DOULIERY / AFP) (Photo by OLIVIER DOULIERY/AFP via Getty Images)

Facebook is still struggling to rein in COVID-19 misinformation in Spanish, a recent report from the advocacy group Avaaz indicates.

Avaaz, a global human rights group, looked at top-performing posts on Instagram, which is owned by Facebook, promoting a debunked viral claim the coronavirus will make patients’ arms magnetic because it contains metals or possibly a microchip. Of the 30 Instagram posts in Spanish, only one had received a fact-checking label as of June 14. In comparison, roughly half of the 47 high-performing English posts had a fact-checking label.

One post flagged by the group showed a woman placing a magnet on her arm, claiming that it stuck to the site where she received her vaccine injection. The post, which racked up tens of thousands of views, has been removed by Facebook.

Facebook employs third-party organizations to fact-check its content. If a post receives a fact-checking label as false, it’s deprioritized in the user’s feed. Identical content also receives the label, including in Spanish.

Advertisement

The research underscores ongoing concerns from lawmakers and advocacy groups that Facebook’s moderation of Spanish-language disinformation trail its efforts to take on the same content in English.

“It should be well known by now that we can no longer count on social media platforms to regulate themselves and protect users against harmful anti-vax lies,” Rebecca Lenn, senior advisor at Avaaz said in a statement. “President Biden, his administration, and Congress [should] make combating disinformation and regulating the tech platforms a priority in ongoing COVID-19 response efforts.”

A 2020 Avaaz report described a similar discrepancy in moderation for COVID-19 misinformation in Spanish. The analysis also indicates that Italian and Spanish-speaking users may be at greater risk of misinformation exposure.

Advocacy groups and lawmakers have also expressed concerns about Spanish-language disinformation surrounding the 2020 election and the coronavirus pandemic.

A campaign led by Free Press, National Hispanic Media Coalition and the Center for American Process launched earlier this year is demanding that the company hire an executive position to oversee U.S. Spanish-language content moderation and publicly explain how its algorithms are trained to audit and report Spanish-language posts and activity.

Advertisement

“These types of findings aren’t surprising,” says Carmen Scurato, senior policy counsel at Free Press. “It’s disappointing because we have seen from Facebook in particular incompetent oversight when it comes to Spanish language on their platform and allowing disinformation to spread to Latinix communities.”

Scurato says that while the company has made modest improvements in the past year, it’s imperative that Facebook take down harmful content in Spanish at the same rates it does in English.

“Those are the issues we’ve been flagging to Facebook for close to a year and we haven’t gotten any clear answers because there’s no transparency,” she said.

Facebook removed 11 of the Spanish-language Instagram posts and 30 of the English-language Instagram posts flagged by Avaaz after CyberScoop contacted the company on June 17. A Facebook spokesperson said some of the posts flagged by Avaaz debunked false claims about the vaccine, which the platform allows.

Facebook said it uses the same machine learning technology to police misinformation in Spanish as English but declined to further comment on the size of its content moderation team for Spanish language posts.

Advertisement

“While we aim to take action against anything that breaks our rules, this report only looked at roughly 75 pieces of content when we’ve removed more than 18 million pieces of COVID and vaccine misinformation and labeled 167 million pieces of this content with fact-checks from our independent partners,” a Facebook spokesperson said.

Facebook said it uses the same machine learning technology to police misinformation in Spanish as English but declined to further comment on the size of its content moderation team for Spanish language posts.

Tonya Riley

Written by Tonya Riley

Tonya Riley covers privacy, surveillance and cryptocurrency for CyberScoop News. She previously wrote the Cybersecurity 202 newsletter for The Washington Post and before that worked as a fellow at Mother Jones magazine. Her work has appeared in Wired, CNBC, Esquire and other outlets. She received a BA in history from Brown University. You can reach Tonya with sensitive tips on Signal at 202-643-0931. PR pitches to Signal will be ignored and should be sent via email.

Latest Podcasts