Tumblr is currently facing the challenge of managing automated flagging and takedowns that have gone awry, causing frustration among its users. Recently, many on the platform have noticed their content being flagged as “mature,” even when it clearly isn’t. This misclassification has affected the visibility of posts, as many users keep their settings configured to hide mature content.
Users have reported a wide array of false flags, from innocent cat GIFs and fandom content to harmless art and even images of hands. The root of the issue seems to point to AI-based automation, a problem not unique to Tumblr. Other platforms like Pinterest and Instagram have similarly faced accusations regarding automation mishaps, although Pinterest attributed their issues to internal errors rather than AI.
The recent problem on Tumblr is linked to an update in its Android app, where the company is refining its mature content filtering systems. The introduction of a new moderation layer to Content Labels is currently being tested, according to a Tumblr spokesperson. The aim is to make Tumblr safer while accommodating diverse interests, with settings adjustable by users.
Acknowledging the issue, Tumblr has been transparent on its blog about the “incorrect classification issues” and is actively working to minimize them. They are also updating the appeal process to handle more cases efficiently.
The exact cause of these false flagging instances remains unclear, and while Tumblr hasn’t specifically pointed to AI, it’s a likely factor. Additionally, reduced staffing after its acquisition by Automattic in 2019 may have contributed. Automattic has since moved Tumblr’s backend to WordPress to improve management and address financial challenges.
It remains to be seen how Tumblr will adjust and evolve its systems to restore user trust and ensure smoother moderation moving forward.
