X tests comment downvotes for Premium users to combat AI spam

X is testing a comment downvote feature exclusively for Premium subscribers as part of an ongoing effort to reduce AI-generated spam and improve reply quality.

Nikita Bier, Head of Product at X, recently previewed the update, bluntly calling the current reply algorithm “the worst product in the company.” He noted that it lacks meaningful signals and often prioritizes low-quality content, such as cryptocurrency spam.

A new thumbs-down icon will appear in the reply function bar. Tapping it opens a menu with five reasons for the downvote: Not Interested in This Post, Incorrect or Misleading, AI Generated, Spam, or Report Post.

Unlike Reddit’s system, these downvotes will not directly hide a reply, alter like counts, or influence broader trending content. Bier explained that the primary goal is to gather user preference data to refine personalized reply rankings and streamline content reporting. The update aims to give paying users more control over their feeds while feeding valuable data back into X’s algorithm.

To prevent coordinated attacks and mass downvoting, the feature is restricted to X Premium subscribers—roughly 1% of the user base. This paywall also helps block bot manipulation, a growing concern as AI tools enable highly sophisticated automated accounts.

While X has tested downvotes before, previous trials focused on reporting rather than ranking. This iteration is heavily data-driven, seeking to improve reply quality through aggregated feedback. If successful, it could expand into a broader system or remain a refined, exclusive perk for paying users.