Tumblr is the latest technology company that is struggling with automated flagning and removal, which went crazy and raised the anger of users.
In recent days, Tumblr users have complained that their content is marked as “mature”, even if this is not the case. The problem reduced the visibility of user posts, because many people on the platform configured their settings to hide mature content by default.
According to many posts of the affected Tumblr users, the posts were falsely marked, even though they are not sexual or violent, and these false removal contained everything with CAT gifs to the fandom content Down piece and even photo of the hands. Some suspect that automation based on artificial intelligence may be guilty of problems.
Similar accusations have harassed other social platforms in recent weeks. For example, Pinterest finally admitted that the internal error led to the bans of the mass user. Meanwhile, Instagram this week refused to comment on its own problems with mass bans, which according to users received a small relationship or attention outside of online complaints on applications such as X and Reddit.
In both cases, users suspected that the fault based on artificial intelligence, although Pinterest denied that it was so.
In Tumblr, the problem with flagning is related to the update of the Android application, in which the company said that it experimenting with the improvements of mature content filtering systems. In particular, he tested a new layer of moderation Content labelsThe company said TechCrunch.
Tumblr spokesman said that the experiments are continued and on the basis of user feedback will be improved before the changes are introduced to other platforms.
“When we work to make Tumblr a safer place for all users, we try to respect a diverse set of interests and preferences of content that can be adjusted in the settings. We consider it a continuous process when we still adapt the method of detecting and solving mature content,” said Tumblr spokesman in e-mail.
Company also confirmed the problem on his blogAnswering the user's question about falsely marked posts. The Tumblr team noticed that they were aware of “incorrect classification problems” and actively worked on their reduction.
In addition, Post explained that the Tumblr cancellation process was updated in the coming weeks to be able to deal with more cases. (Tumblr did not answer our questions about planned changes in the appeal process.)
Whether AI's fault or other automation is clear, because Tumblr would not specifically talk about the cause of new problems. However, reduced staff in the blogging service probably also played a role.
After taking over in 2019 by WordPress.com Maker Automattic, Tumblr faced dismissals because his employees were again assigned to other projects in the dominant company. Last year, Automattic announced that Tumblr's facilities would also be transferred to WordPress to facilitate management and stop its financial losses.
