LinkedIn will stop recommending generic, AI-written posts to anyone beyond the author’s own connections, a change the Microsoft-owned platform set out in a company blog post on 18 May and that its new chief executive, Daniel Shapero, has personally put his name to in a video to members.

The move lands as professional feeds fill with AI-generated filler that reads as polished but says little, and as platforms from YouTube to TikTok tighten their own rules on low-effort synthetic content. For marketing managers who have spent a year being told to “do something with AI”, it draws a sharper line than most internal guidance has managed: the platform will keep rewarding a genuine point of view and quietly bury the rest.

What exactly is LinkedIn changing?

LinkedIn will limit the reach of posts and comments its systems judge to be generic and AI-generated, rather than deleting them. Flagged content stays visible to an author’s direct connections but is held back from the feed recommendations that drive most of a post’s reach. Vice-president of product Laura Lorenzetti, who set out the changes, said the detection correctly identified generic content in 94% of early tests — a figure the company has not broken down, and one published without any data on how often genuine posts are caught by mistake.

The system targets three things: generic AI posts and comments, engagement-bait video, and the automation tools that generate content in bulk. Lorenzetti named specific tells, including recycled “thought leadership” and the “it’s not X, it’s Y” construction that has become a signature of machine-written prose. AI-assisted writing remains allowed; the line LinkedIn is drawing falls between using a model to sharpen a real idea and using one to manufacture a post with no idea in it. Content creation on the platform is up 14% year on year, Lorenzetti said — the growth that made the problem worth policing. The company has said the rollout will be gradual, and that it may be several months before the effect is visible across feeds.

Why has the new chief executive put his name to it?

Daniel Shapero, who became LinkedIn’s chief executive on 22 April after seven years as chief operating officer, used a video message to frame the clean-up as a direct response to member feedback. He described building classifiers to identify content that does not “represent a unique point of view from the person”, and using them to limit how far that content spreads beyond someone’s network — with the same approach applied to comments that only summarise a post rather than add to it.

Fronting the change with a brand-new chief executive is a signal in its own right, and it sits awkwardly against the rest of the picture. Microsoft, which owns LinkedIn, has continued to promote AI writing across its products, in one recent demo using a generic LinkedIn post to show off a Copilot feature. The policy also arrives in the same stretch as job cuts: more than 600 roles, concentrated in California, were reported in May, which outside estimates put at roughly 5% of LinkedIn’s global workforce — a figure the company has not confirmed. In an internal memo reported by Business Insider, Shapero wrote that LinkedIn needed to “reinvent how we work” and make “hard trade-offs”, but did not name AI as the cause, and the company has not tied the cuts to its AI strategy. LinkedIn’s revenue rose 11% in its latest quarter.

What does it change for marketers?

Marketers who already publish from real experience have little to fear and something to gain from a change that tilts distribution towards exactly the work they do. The practical reading is narrow and useful: the platform is not penalising AI, it is penalising the absence of a point of view, so a model used to tighten a genuine argument is safe while a model used to spin up posts at volume is the target. For teams under pressure to keep up a posting cadence, that reframes the task from producing more to producing something only they could say.

It also recalibrates the comment, which has quietly become a reach tactic of its own. Generic AI replies that restate the original post will now be held back, while a short, specific response that adds a fact or a dissent is the kind of contribution the system is built to reward. None of this asks marketers to post less or to drop the tools; it asks them to keep their own judgement in the loop — the part LinkedIn says it is trying to surface, and the part no classifier can supply on a marketer’s behalf.

LinkedIn has not said when the system will be fully live, nor released the false-positive data that would show how often real posts are mistaken for slop. For now the line between an AI-assisted post and an AI-generated one is being drawn by a classifier the company has described but not opened up, with the practical test playing out in feeds over the coming months.