COVID-19 affected Facebook’s ability to remove harmful and forbidden material from its platforms, the social media giant says.
Sending its content moderators to work from home in March amid the pandemic led the company to remove less harmful material from Facebook and Instagram around suicide, self-injury, child nudity and sexual exploitation.
Sending its human reviewers home meant it relied more on technology than people to find posts, photos and other content that violates its rules.
“Today’s report shows the impact of COVID-19 on our content moderation and demonstrates that, while our technology for identifying and removing violating content is improving, there will continue to be areas where we rely on people to both review content and train our technology,” vice president Guy Rosen wrote in a blog post.
The company said on Tuesday it has since brought many reviewers back to working online from home and, “where it is safe,” a smaller number into offices.
But Facebook also said its systems have gotten better at proactively detecting hate speech, meaning it is found and removed before anyone sees it.
The company said its detection rate increased 6 points in the second quarter, to 95 per cent from 89 per cent.
Facebook said it took action on 22.5 million pieces of content – like posts, photos or videos – for hate speech violations in the second quarter, up from 9.6 million in the first quarter.
The social network said that’s because it has expanded its automation technology into Spanish, Arabic and Indonesian and made improvements to its English detection technology.
Facebook also announced Tuesday that it is banning caricatures of Black people in the form of blackface, as well as dehumanizing depictions of Jewish people that include images or other depictions of Jewish people running the world or controlling major institutions such as media networks, the economy or the government.