Every so often a platform ships a feature that functions as a confession. LinkedIn's new feedback option — a button, attached to posts and comments, reading "seems like AI slop" — is one of them. A network does not build a complaint mechanism for a problem it does not have.
The scale of the problem is the interesting part. Third-party analyses cited by the Wall Street Journal estimated that 41% of long-form posts on the network were AI-generated in one sample, along with 30% of all comments; another sample put 81% of posts as likely using AI more than moderately. LinkedIn disputes the vendors' methods, and some scepticism is fair — AI detection is a famously unreliable trade. But the company's own countermeasures tell you what it believes: it says it blocks more than 200,000 AI-generated spam comments every day, and that its systems already withhold an estimated 94% of generic or AI-generated posts from travelling beyond the author's immediate network. You do not build a dam that size across a dry riverbed.
For the member — there are more than a billion — the changes cash out as follows. The slop button lets you flag the worst of what reaches you, and feeds the classifiers deciding what reaches everyone else. The "enhance your post" feature, which invited the machine in, is being removed in favour of a plain proofreading tool. And a private "nudge" is coming that will tell creators when a post reads as inauthentic — the algorithmic equivalent of a colleague taking you aside.
There is an irony here worth savouring for exactly one paragraph. LinkedIn spent years embedding AI into recruiting, search and writing tools; it sold the pen, and is now installing the complaints desk. "How can we make sure that your time on LinkedIn is a good investment," is how Dan Roth, the network's editor in chief, frames the turn. The honest answer is that the investment case collapsed the moment members began suspecting that neither the post nor the applause beneath it involved a human being.
The practical takeaway, though, sits in that 94% figure, and it applies to you. If LinkedIn's classifiers decide your writing reads as generic — whether a machine wrote it or you simply write like one — your post quietly goes nowhere beyond your own connections. The throttling is already on. Which makes the strategy for being read on LinkedIn in 2026 unusually simple, and rather cheering: sound like a person. Specifics, opinions, the occasional imperfect sentence. The machines flooded the feed with polish, and polish is now the tell. Write like yourself; it has become a distribution strategy.
