Facebook has recently released a report that detailed how well the social media giant enforced its community standards. This report indicated the total number of content that violates the Facebook’s Community Standards, and how well the social media platform was able to detect and remove content using its detection technologies. In its report, the company focused on six key areas, which are graphic violence, adult nudity and sexual activity, terrorist propaganda, hate speech, spam, and fake accounts.
The social media giant admitted in its report that it still needs to improve its ability to more quickly detect and remove hate speech from its platform. In the first quarter of 2018, only 38-percent out of 2.5 million pieces of hate speech were identified by the company's detection technology, although it should be noted that this figure is significantly better compared to the last quarter of 2017, where only around 24-percent of 1.6 million pieces of hate speech were detected by Facebook's technologies. To enhance its capacity to detect hate speech, Facebook recently rolled out a new hate speech checker to its users, which could help in improving the company’s AI-based detection of hate speech.
The social media giant also claimed in its report that it was able to remove around 837 million pieces of spam content and 583 million fake accounts in the first three months of 2018. In contrast to its ability to detect hate speech, almost all of the spam content and fake accounts were removed before any user reported the violating content and accounts, although the total number of fake accounts detected by the company in the first quarter of 2018 were lower compared to the last three months of 2017. Meanwhile, around 21 million pieces of content that contain adult nudity and sexual activity have been removed by Facebook in the first quarter of 2018, while improvements made to its photo-matching technology allowed the social media giant to detect 3.5 million pieces of content that depict graphic violence, which is considerably higher than the 1.2 million pieces of violating content that was detected in the previous quarter. Overall, improvements in technology allowed the social media giant to combat spam, fake accounts, and other forms of violating content more quickly, although the company admitted that there is still room for improvement in detecting offensive content especially in posts that contain less commonly used languages due to the lack of data needed to train its algorithms.