Tumblr Users Say AI Is Mistakenly Labeling Their Posts as “Mature”

Tumblr is the latest tech platform facing backlash over malfunctioning automated systems that flag and remove content.

In recent days, users have reported that Tumblr is incorrectly labeling harmless posts as “mature,” limiting their visibility since many users have default settings that hide such content.

Harmless Posts Wrongly Flagged as “Mature” spark AI Moderation Concerns

Numerous users have complained that the system wrongly flagged non-explicit posts — including cat GIFs, fan art, and even picture of hands. Many believe faulty AI moderation tools may be behind the errors.

Other social media platforms have faced similar criticism. Pinterest recently acknowledged that an internal mistake led to a wave of user bans. Meanwhile, Instagram has ignored growing user frustration as people flock to online forums like Reddit and X to discuss widespread account suspensions.

In both instances, users believed AI-driven moderation was responsible, although Pinterest rejected that explanation.

On Tumblr, the flagging problem appears linked to a recent Android app update. The company explained that it has been testing enhancements to its mature content filtering, including a new moderation layer added to its Content Labels system, according to comments made to TechCrunch.

Tumblr Says Tests and User Feedback Will Shape Updates Before Wider Rollout

A Tumblr spokesperson said the team is still testing the changes and will use user feedback to guide improvements before rolling out the update to other platforms.

A Tumblr spokesperson said via email that the company is working to make the platform safer for everyone while respecting diverse interests and content preferences, which users can customize in their settings. “This remains an ongoing effort as we refine our methods for identifying and managing mature content.”

The company also addressed the issue on its blog in response to a user inquiry about incorrectly flagged posts. Tumblr’s team acknowledged the problem, stating they were aware of the “misclassification issues” and were actively working to minimize them.

Tumblr To Update Appeals Process Amid Moderation Issues, With Causes Still Unclear

The blog post also mentioned that Tumblr plans to update its appeals process in the coming weeks to better manage a larger volume of cases. (Tumblr declined to comment on the specific updates to the appeals process.)

It’s unclear whether AI or other forms of automation are responsible for the recent issues, as Tumblr has not provided specific details about the root cause. However, the platform’s reduced workforce likely contributed to the problem.

Since Automattic — the company behind WordPress.com — acquired Tumblr in 2019, the service has experienced staff reductions, with employees being reassigned to other projects within the parent company. In an effort to cut losses and simplify operations, Automattic also announced last year that Tumblr’s back end would be migrated to the WordPress infrastructure.

Read the original article on: TechCrunch

Read more: Google Plans To Expand AI Fraud Detection And Security In India