Facebook Removed Nearly 40% More Terrorist Content in Second Quarter

Facebook removed about 8.7 million pieces of content it categorizes as terrorism in the second quarter.

Photo: josh edelson/Agence France-Presse/Getty Images

Facebook Inc. FB -2.61% removed nearly 40% more content that it categorized as terrorism in the second quarter compared with the first three months of the year, the company said.

Facebook removed about 8.7 million pieces of such content—which includes, according to the company’s definition, nonstate actors that engage in or advocate for violence to achieve political, religious or ideological aims—in the second quarter of this year, up from 6.3 million in the first quarter.


Where and how should Facebook draw the line on extremist speech? Join the conversation below.

That increase “was largely driven by improvements in our proactive detection technology to identify content that violates [Facebook’s] policy,” a company spokeswoman said.

For “organized hate” groups, a separate category, the company said it took down four million pieces of content, down from 4.7 million in the first quarter.

And in a third category, hate speech, the company said it dramatically increased removals of such posts from Facebook and its Instagram platform. It said it removed 3.3 million pieces of hate speech from Instagram in the second quarter, more than triple the amount the previous quarter, and 22.5 million pieces from Facebook, more than double what it removed in the first quarter. The company credited its technology’s expansion into several languages, including Spanish, Arabic and Indonesian, for the increase.

Instagram removed about 388,800 pieces of terrorist content, down from 440,600 in the first quarter, but it removed more organized hate content in the second quarter—266,000 pieces versus 175,100 in the previous quarter.

Read More

  • States Call on Facebook to Launch Hate-Speech Hotline (Aug. 5, 2020)
  • Disney Slashed Ad Spending on Facebook Amid Growing Boycott (July 18, 2020)
  • Why Some Hate Speech Continues to Elude Facebook’s AI Machinery (July 9, 2020)

White supremacist groups have been a focus for the social-media giant. Since October of last year, the company said it completed 14 network takedowns to remove 23 organizations in violation of Facebook’s policies. The majority of those takedowns, nine of the 14, targeted “hate and/or white supremacist groups,” including the KKK, the Proud Boys, Blood & Honour and Atomwaffen, the company said.

Facebook has put more than 250 white supremacist groups on its list of dangerous organizations, company officials earlier said, putting them alongside jihadist organizations such as al Qaeda.

The tech giant also recently banned a large segment of the boogaloo movement from its platform. Adherents of the loosely-organized boogaloos include white supremacists.

Facebook removed the boogaloo accounts after a targeted investigation by human analysts, officials said. The company has increasingly turned to humans to assess networks that actively try to avoid its automated content-monitoring tools.

Facebook’s latest figures come after the company said in June it took down posts and ads for President Trump’s re-election campaign because they violated the company’s policy against “organized hate.” The campaign ads claimed that “Dangerous MOBS of far-left groups” are causing mayhem and destroying cities and called on supporters to back Mr. Trump’s battle against antifa, a loosely organized activist movement that the White House has blamed for unrest.

The ads featured a large, red downward-pointing triangle. The inverted red triangle is a marking Nazis used to designate political prisoners in concentration camps, according to the Anti-Defamation League and other groups. The Trump campaign said that the triangle is a common antifa symbol, though some experts on extremist groups have disputed that.

Other social-media platforms have also faced pressure to remove information tied to groups classified as terrorist organizations. Last year, Twitter Inc. suspended accounts linked to Palestinian group Hamas and Iran-backed militant group Hezbollah. U.S. lawmakers had criticized Twitter for allowing those entities to remain active on the platform even though the State Department designated both as terrorist organizations.

Social-media companies have also faced tensions over how to handle misleading posts. Twitter earlier this year labeled tweets by Mr. Trump about mail-in ballots as misinformation, highlighting a widening divide among big tech platforms on how they handle political speech as the U.S. presidential election approaches.

Write to Rachael Levy at [email protected]

Copyright ©2020 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8