Under every post on LinkedIn, a new option “Seems like AI slop” has appeared in the menu. Anyone whose feed shows your post can use it to signal to the platform that they consider the content machine-generated. A competitor, a client or a former colleague. The penalty for such a flag is quiet. Flagged content gets lower reach, with no warning and no visible trace.
LinkedIn's product director, Hari Srinivasan, described in an update a few weeks after launch how many people use the button. More than a million people have flagged content as AI slop. According to Srinivasan, material that the platform itself classifies as AI slop gets 40% fewer views.
How the button works
LinkedIn launched the feature on 30 July. It sits in the menu under the three dots on each post and works as a private signal, not a public report. The platform dampens flagged content in recommendations outside your own network, which is exactly where companies and creators look for new audiences. Srinivasan frames AI slop as a priority. “AI slop is a top priority for all of us. We really care about this,” he wrote in the update.
The other side of the feature is feedback for the author. Anyone who gets a signal from the community that their post seems inauthentic or too machine-like sees a private notice in Post Analytics. No public label, no appeal, just a quiet note in the overview. For a brand that builds the personal profiles of its managers on LinkedIn, this is a new kind of metric. Not how many people saw a post, but how many people it came across to as artificial.
On the most common objection, that the button could be abused to target a competitor, Srinivasan says the platform has safeguards to stop individual feedback from unfairly singling out a particular person. He has not yet published the details of those safeguards, so how strong the protection really is cannot be verified from outside.
A feed flooded with machine text
The scale of the problem was described by a study from detection firm Pangram, which scanned more than a million posts across five platforms. LinkedIn came out as the most flooded with machine text. In posts longer than 250 words, Pangram classed more than 40% as fully AI-generated.
LinkedIn is also fighting a side that users do not see. According to Srinivasan, the platform blocks hundreds of thousands of attempts at automated slop comments every day. In the same move, LinkedIn retired its own “enhance your post” tool, which helped users write their texts, and replaced it with a proofreader that, according to the company, fixes the language but does not change the author's voice. The platform is drawing a line between AI that smooths out a human text and AI that replaces it outright.
Two levels in one feed
The feed is becoming a two-level space. The signal that decides reach is no longer whether a machine helped with the writing, but whether the audience reads the result as human. The difference is practical. A manager who has a model generate a whole post and publishes it unchanged risks readers sending it into muted reach with one click. Someone who uses the model for an outline and an edit, and then rewrites the text in their own voice, passes through the same gate as before.
Creator economy analyst Lindsey Gamble, who summarised the update, estimates that the feature will go through several more iterations and that platforms will eventually let users filter between human content, AI content and mixes of the two. If that happens, reach will stop being purely a matter of the algorithm and become a choice for the reader. For every brand that bases its LinkedIn content on generative tools, reach becomes a variable that the crowd partly controls.
Comments