LinkedIn admits its feed has become an AI-slop dump and launches a ‘Looks like AI slop’ button

A study from Originality.ai published last month lays out the scale of the problem: 81% of long-form posts on the platform — those crossing the hundred-word threshold — were generated with AI. That is a sharp jump from just 50% in 2025. The number surprises no one who has scrolled the feed lately, but it forced leadership to stop ignoring the issue and start acting.

The immediate fix is a new report button tucked into the hamburger menu on every post. Label: ‘Looks like AI slop’. Clicking it removes the post from the reporter's feed and routes it into a training pool that will sharpen the automated filters. Harry Srinivasan, head of product, wrote in a post of his own that "AI slop is a top priority for all of us," adding that users come to the platform to connect with real people and real insights, not generic text masquerading as management wisdom.

Alongside the report button, LinkedIn is shutting off a feature that until now let users generate posts directly inside the interface. Going forward, the built-in AI tools will be limited to proofreading only — no added superlatives, no forced insights, no "lesson I learned" framing. It is a move that acknowledges the platform itself supplied ammunition for the flood, and is now trying to gather up the spent rounds. A move that finally admits LinkedIn itself helped load the gunpowder for the flood.

Srinivasan said the company is also building a mechanism to notify creators when their posts are flagged as inauthentic, "in the easiest and least offensive way possible." It remains unclear whether that means a private notice, an algorithmic reach reduction, or a visible label. What is certain: the platform is trying to walk a tightrope between cleaning up the feed and scaring off legitimate users who rely on AI for editing.