LinkedIn announced its "Seems like AI slop" button, reachable from the three dots menu on a post, on July 30. According to chief product officer Hari Srinivasan, "over a million people" have now clicked it.

Why the button exists

The timing was not coincidental. LinkedIn announced the button a few weeks after the AI detector Pangram determined that 41 percent of the platform's longform posts were flagged as fully AI generated, a finding 404 Media reported.

The platform made other changes alongside it:

  • New classifiers — "new and improved" models to identify posts as AI generated.
  • A removed feature — the tool that would "enhance your post" with AI was taken away.
  • Comment enforcement — the company had already said it would crack down on comments created at scale with little or no human involvement.

The third item is worth noting: LinkedIn is not only flagging AI content but withdrawing the AI tool it built itself. Encouraging users to generate text while complaining about the result was an incoherent position.

What the number means

Srinivasan says users are now seeing 40 percent fewer views on what the platform classifies as AI slop. The report button is not merely a feedback box, then; it feeds the ranking algorithm directly.

The platform is also adding a notice aimed at the author: whoever posted will see a message saying "Some members told us this post seems like AI." Srinivasan framed it this way: "We approached this assuming good intent; I know I'm increasingly conscious on how to not sound like AI & the goal is to provide helpful feedback."

The part that is not measured

The figures need careful reading. A million clicks does not mean a million distinct slop posts; many people may have clicked the same one. More importantly, the button rests on human judgment, and people are not reliably good at recognizing AI text. Whether a post was actually written by a model and whether members feel that it was are not the same thing.

This shows in the platform's own framing: the company says "what we classify as AI slop," not "written by AI." What gets measured is not the origin of the content but the impression it leaves. A well-written AI post can pass the filter; a carelessly written human post can trip it.

The direction still means something. A platform enlisting its own user base in moderation and feeding the result into ranking may be the only approach that scales against an abundance of machine-written text. The alternative is running every post through a classifier and living with that classifier's mistakes.