×
You will be redirected back to your article in seconds

Facebook Touts Automated Takedowns of Porn, Fake Accounts — but Says Hate-Speech Filters Need Work

Facebook has published, for the first time, a report with internal statistics about its enforcement efforts to combat violations of its content standards.

According to the numbers, covering the six-month period from October 2017 to March 2018, Facebook’s automated systems remove millions of pieces of spam, pornography, graphic violence and fake accounts quickly — but that hate-speech content, including terrorist propaganda, still requires extensive manual review to identify.

Facebook’s new Community Standards Enforcement Report “is very much a work in progress and we will likely improve our methodology over time,” Chris Sonderby, VP and deputy general counsel, wrote in a blog post about the report.

The new disclosures from Facebook are part of its larger effort to rebuild trust among users — and advertisers — after widespread concern from lawmakers and regulators about its content-management practices. In addition, Facebook remains in damage-control mode after revelations about how political-consulting firm Cambridge Analytica improperly obtained data on millions of users without their knowledge and consent.

Key statistics from the report:

  • Facebook disabled about 583 million fake accounts in Q1, most of which “were disabled within minutes of registration.” Overall, the social giant estimated that around 3%-4% of active Facebook accounts on the site during Q1 were still fake.
  • The company took down 837 million pieces of spam in Q1 2018, nearly all of which was flagged before any users reported it.
  • Facebook removed 21 million pieces of adult nudity and sexual activity in Q1 2018, 96% of which was found and flagged by its technology before it was reported.
  • Facebook took down or applied warnings labels to about 3.5 million pieces of violent content in Q1 2018 — 86% of which was identified by our technology before it was reported to Facebook.
  • For hate speech, “our technology still doesn’t work that well and so it needs to be checked by our review teams,” VP of product management Guy Rosen wrote in a separate post. Facebook removed 2.5 million pieces of hate speech in Q1 2018, of which just 38% was flagged by automated systems.

Facebook’s latest transparency report also includes data on copyright and trademark infringement reports. During the six-month period, Facebook and Instagram took down 2.78 million pieces of content based on 373,934 copyright reports, as well as 222,226 pieces of content based on 61,172 trademark reports and 459,176 pieces of content based on 28,680 counterfeit reports.

Meanwhile, Facebook in Q1 2018 flagged 1.9 million terrorism-related posts, from related to ISIS, al-Qaeda and affiliated groups, up from 1.1 million in Q4 2017. It attributed the increase to “improvements in our ability to find violating content using photo-detection technology, which detects both old content and newly posted content.”

Separately, in an action more specifically related to the Cambridge Analytica user-data scandal, Facebook on Monday said it suspended 200 third-party apps that had access to large amounts of user info pending review of whether they misused that data.

More Digital

  • Variety Cord Cutting Placeholder Cable

    Cord Cutting Will Accelerate in 2019, Skinny Bundles Poised to Fail (Report)

    The pace of cord cutting is continuing to accelerate this year, according to a new Convergence Research Group report, with 4.56 million TV households opting to ditch pay TV. By the end of the year, 34% of U.S. households won’t have a traditional TV subscription, according to the research company’s latest “Battle for the American [...]

  • Missing Link Laika Studios

    ‘Missing Link’ Again Tops Studios’ TV Ad Spending

    In this week’s edition of the Variety Movie Commercial Tracker, powered by the TV ad measurement and attribution company iSpot.tv, Annapurna Pictures claims the top spot in spending for the second week in a row with “Missing Link.” Ads placed for the animated film had an estimated media value of $5.91 million through Sunday for [...]

  • Sirius Logo

    SiriusXM Unveils $8 Essential Plan for Consumers Without Cars

    SiriusXM wants to cater consumers without cars, or cars without compatible stereos, with a new $8 plan for mobile and in-home listening. Dubbed SiriusXM Essential, the plan offers access to 200+ channels featuring the network’s entire music programming, as well comedy, news and select sports channels. Consumers will be able to test the new plan [...]

  • Mueller Report Book Editions Top Amazon's

    Mueller Report Book Editions Shoot to Top of Best-Seller Lists at Amazon, Barnes & Noble

    Robert Mueller is now a best-selling author. Book publishers’ forthcoming editions of the special counsel’s report zoomed to the top of the Amazon’s and Barnes & Noble’s lists of book best-sellers Friday. That comes a day after the report was publicly released, culminating the nearly two-year investigation into Russian interference in the 2016 U.S. election [...]

  • Marques Brownlee - Retro Tech

    YouTube Orders Marques Brownlee 'Retro Tech' Original Series

    YouTube has turned to one of its homegrown stars — technology vlogger Marques Brownlee, aka “MKBHD” — for its newest original series. The video platform has greenlit series “Retro Tech” featuring Brownlee, in which he’ll unbox and review vintage technology products that have defined pop culture. The show, slated to debut in December 2019, follows [...]

  • Netflix Tests Random Episode Button in

    Netflix Starts Testing Random Episode Button

    Netflix is testing a button to play random episodes of select TV shows, the company confirmed Friday morning. “We are testing the ability for members to play a random episode from different TV series on the Android mobile app,” a Netflix spokesperson told Variety. “These tests typically vary in length of time and by region, [...]

More From Our Brands

Access exclusive content