TikTok, removed 81 million videos in three months

TikTok: i video da 5 minuti potrebbero essere alle porte thumbnail

Lately TikTok announced an update on content moderation on the platform, which transparently reveals the data for April, May and June of this year. In these three months, he has removed 81,518,334 videos. And TikTok puts this number in proportion by saying that i removed videos are 1% of the total, which therefore hovers around the eight billion, around 90 million videos a day.

TikTok, 81 million videos removed in three months

In July, TikTok announced that it has implemented a technology that automatically removes all content that violates the guidelines. Problematic videos for the health of minors, the nudity and the sexual activity, i violent content and the illegal activities. The app has automated the removal of this content because they ensure that their AI algorithms have the greatest accuracy in these areas. But the social also ensures that it has a number of moderators in the flesh who check the reported content.

Upon announcement, TikTok explained that the percentage of false positives is around 5%. If this is also true for the reported data (when it appears that TikTok was just testing the automatic removal), it means that about 4.6 million of videos were brought back online after human checks.

According to TikTok data, 41.3% of content violated child health policies. 94.1% of videos were removed before anyone could see it. 27.518 video were removed for reporting fake news on COVID-19. 87.6% of videos would be removed within the first 24 hours, 69.7% got zero views. Conversely, information from the WHO and the ministries of health around the world has reached 921 million views.

After reaching one billion active users last month, TikTok continues to grow. It seems that the ByteDance social network is intent on communicate transparently how it monitors content on the platform. But the experience of these years of social networks teaches that the question is complex. Software control can go wrong and managing such a huge (and multi-lingual) network is difficult. But it seems there is more and more interest to address this generational problem.