What the button produces is an input, not a decision
The wording of the option is specific: a post is reported as seeming like AI slop. According to a post by LinkedIn's chief product officer, Hari Srinivasan, these reports feed as signal into the company's models to reduce low-quality content in the recommended feed. The company also says it blocks hundreds of thousands of automated comment attempts a day and has strengthened its classifiers.[1]
A reader saying "this looks generated" is an unverified guess about where the post came from. When the system feeds that guess into a ranking model, an unverified judgment about origin turns into weight on visibility. The competing account is that reports are used as one weak feature among hundreds, in which case individual errors matter little. The announcement gives nothing that would let a reader choose between the two.[1]
The tool moving from generation to correction
In the same announcement the platform is also retiring its own text-generation tool. The enhancement feature that helped compose posts is being replaced by a tool that corrects existing writing without changing the author's voice.[1]
The two moves touch the same object from opposite ends. The platform is trimming the volume of synthetic text it produces itself while handing the flagging of the rest to users. Shifting the burden of establishing origin from the producer to the reader does not answer the question of who knows where the text in the feed came from; it relocates it.[1]
The three documents that would close the gap
Three things are absent from the announcement: the threshold at which a report affects visibility, the rate of mistaken reports, and whether an appeal route exists for a post whose reach is unfairly reduced. When and where the features roll out is unwritten too. Once those three line items are published, whether the system protects readers becomes testable from outside for the first time; until then what exists is the wording on a button.[1]