Facebook's Algorithms Undermined COVID-19 Response, Spread Violent Misinformation After George Floyd's Killing

An Accountable Tech study finds that Facebook’s content curation and recommendation algorithms have actively promoted the spread of dangerous misinformation and extremism.

The United States has faced two major inflection points this year: the virulent spread of COVID-19 and the fight for racial equity. In both instances, Facebook’s content curation and recommendation algorithms have actively promoted the spread of dangerous misinformation and extremism, directly threatening public health and safety.

An Accountable Tech study of the platform in the months following the COVID-19 outbreak and the killing of George Floyd found that Facebook’s algorithms pushed users to join Groups, “like” Pages, and engage with content that have promoted COVID-19 conspiracy theories, encouraged violence against protestors, and actively spread racist and hateful speech.

These results are not aberrations. Facebook designed its products to maximize engagement, and that’s exactly what they do. As the company’s own research has made clear, their algorithms inherently boost the most divisive content and steer users to increasingly extreme social and information spheres.

To the extent Facebook is attempting to moderate COVID-19 misinformation and racist content, they are fighting against their own toxic content-shaping algorithms – and they are failing. Accountable Tech’s study of the platform found:

  • Despite Facebook’s claim that Groups in violation of platform policies are removed from algorithmic recommendations, Facebook repeatedly suggested joining Groups falsely touting the efficacy of hydroxychloroquine, spreading dangerous misinformation about testing, and calling for violence against BLM protestors. In promoting these Groups – including “Hydroxychloroquine,” “PLANdemic,” “Black Lives Matter is a Terrorist Organization,” and “In Defense of the Confederacy” – Facebook proactively spread misinformation already marked as false by the platform’s fact-checking partners.
  • After “liking” several ostensibly innocuous Pages – including “FLOTUS Melania” and “Keep Trump in Office” – Facebook’s algorithm recommended users follow a myriad of pandemic and BLM conspiracy theorists, contrary to claims the company no longer recommends such Pages. As a result, by simply heeding Facebook’s prompts, users very quickly end up with a News Feed riddled with COVID-19 misinformation and racist content, including claims that facemasks are “mind control devices,” that COVID-19 vaccines will cause autism, and that George Soros is behind BLM protests.
  • Despite claims by Facebook that the company is promoting authoritative information, viral COVID-19 misinformation and racist content has consistently won out. For example, a video live-streamed by Breitbart News claiming that masks and shutdowns are ineffective in stopping the spread of COVID-19 was viewed more than 20 million times before Facebook removed it – more views than any CDC video posted on Facebook since the pandemic began. Further, four out of five of the highest-performing posts on Facebook about Geroge Floyd have come from the controversial Preacher Franklin Graham, who is regularly featured on the Southern Poverty Law Center’s “hate watch.”