LinkedIn swaps AI writer for proofreader, not a generator

LinkedIn's AI post writer is dead. Per Metricool's blog, the feature that once drafted entire posts from a prompt has been replaced with a proofreading tool, one that will only refine text a user has already written. It will fix grammar, tighten phrasing, suggest a stronger line. It will not conjure a paragraph from nothing. That distinction is the whole story.

This is not a minor UI tweak. For two years, LinkedIn sold the fantasy that a blank text box and a half-formed thought were enough: type "announcing our Q3 results" and the platform would hand back three paragraphs of confident corporate prose. Plenty of brand accounts and executives used exactly that shortcut. The new tool closes the door on it. You now need a draft before LinkedIn will help you improve it. The platform has moved from co-writer to copy editor.

Why the sequence matters more than the single change

Metricool frames this as part of "a run of changes aimed at keeping machine-generated posts out of the feed," and the sequencing is the interesting part. LinkedIn has spent the past year quietly building detection for AI-pattern text, deprioritising posts that read as templated, and pushing its "original" ranking signals harder. The proofreader swap is not an isolated product decision. It is the user-facing admission of a direction LinkedIn's ranking systems have already been moving in. When a platform redesigns its own writing tool to stop generating text wholesale, it is telling you, bluntly, what its algorithm now penalises.

That has a direct cost for any organisation that treated AI drafting as a volume strategy. The playbook of the past eighteen months, batch-generate a month of posts, lightly edit, schedule, worked because the feed could not yet tell the difference. It increasingly can. LinkedIn's own engineering has talked publicly about detecting low-effort and repetitive content patterns; this tool change is the retail version of that enforcement, delivered where users will actually notice it: inside the compose box.

Who this actually hurts

Not the solo creator dashing off a hot take, who never needed a generator in the first place. It hurts institutional accounts: the financial services firm with a four-person social team covering four executives, the multilateral agency pushing out statements in six languages, the industrial group with a content calendar built by an agency and populated by prompt. These are exactly the organisations that adopted AI drafting fastest, because they had the volume pressure and the compliance need for consistent tone. LinkedIn has just made their default workflow harder, not by banning AI outright, but by removing the tool that made mass-production frictionless.

The institutions best placed to adapt are the ones that were already using AI as Metricool's new tool now demands: to sharpen a real draft, not replace one. A policy institution whose comms lead writes a rough 200 words after a board meeting and uses an editing pass to tighten it will sail through unaffected. A comms team that fed "write a LinkedIn post about our sustainability report" into a generator and published the output verbatim will find its next six months of content performing worse, not because LinkedIn banned the words, but because the ranking system was never fooled, it was simply slower to act than the compose box now is.

There is a reputational angle here too, and it is the one that should worry executives more than reach. LinkedIn's authority economy runs on the premise that a named individual, a CFO, a secretary-general, a country director, is saying something they actually think. Generated text, however fluent, erodes that premise the moment a reader suspects it. B2B buyers on LinkedIn, particularly the senior, skeptical ones who follow financial institutions and UN agencies, have gotten reasonably good at spotting the cadence of machine prose: the three-part lists, the em-dash flourishes, the relentless positivity. LinkedIn removing its own generator is a tacit admission that the company agrees with that skepticism.

The practical shift is small and the strategic one is not. Draft first, in your own words, however roughly. Let the tool polish, not populate. Any organisation still running AI-first content production should treat this as the moment to rebuild the workflow around a human first draft, because LinkedIn has just told you, through product design rather than policy memo, which side of that line it intends to keep rewarding.