What happened
On July 30, Business Insider reported that LinkedIn is introducing a feature that lets members flag a feed post that seems like AI slop. The article says the flagged post disappears from that member's feed, and LinkedIn is testing a way to privately tell the author that other members felt the post came across as inauthentic.
The Verge corroborated the rollout the same day and said the option appears in the three-dots menu on posts. The Verge also reported that LinkedIn is ramping up classifiers for AI slop and generally low-quality content, with the goal of reducing that material in suggested posts and content from outside a member's network.
Both reports point back to LinkedIn chief product officer Hari Srinivasan's explanation of the feature. The important distinction is that LinkedIn is not treating AI and slop as the same thing. Business Insider quoted Srinivasan saying the company wants feedback from real people on what sounds authentic, instead of relying only on an AI detector that may get it wrong.
LinkedIn's own June 4 product post sets the broader policy frame. LinkedIn said AI can be a useful tool for refining language, but posts and comments still need to represent the member's voice and perspective. It also said content that appears AI-generated and lacks clear perspective is less likely to be widely distributed beyond the author's immediate network.
What the button can tell you
This is not a clean platform ban on AI-assisted writing. LinkedIn's public position leaves room for AI as a language aid. The reported change is aimed at low-effort, generic, repetitive, or inauthentic professional content.
It is also not proof that any single reader can correctly identify how a post was made. A member can dislike a post for many reasons. A polished human-written post can still be empty. An AI-assisted post can still be useful when it is grounded in real experience and reviewed by the person whose name is on it.
LinkedIn is using the new button as feedback for feed quality, testing private feedback to authors, and replacing its post-enhancement feature with a proofreader that preserves the member's voice, according to The Verge.
REC's read: the problem is source poverty
The weak professional post usually has the same flaw whether AI wrote it or not: there is no visible source of judgment.
It may have a confident lesson, a tidy structure, and a smooth ending. But it does not show the decision, constraint, tradeoff, failure, customer question, research note, product choice, or example that gave the idea weight. Readers can feel that absence even when they cannot prove the workflow behind it.
AI makes this easier to mass-produce because it can assemble a fluent post from a thin prompt. But the deeper problem is source poverty. If the prompt contains no real work, the result has little to protect it from becoming generic. Editing can polish the surface, but it cannot invent accountable experience without crossing into fiction.
For founders, consultants, researchers, operators, and educators, the practical response is to move the source upstream. Do not begin with a prompt asking for a thought leadership post. Begin with a real answer to a real question. What changed your mind? What did the work show? What did you reject? What caveat keeps the claim honest? What example would let a peer inspect the point?
Why recorded answers help
A research-guided video interview gives a team a stronger starting point than a blank posting prompt. The person speaks from their own work. The conversation can ask for examples, evidence, constraints, and decisions. The transcript then becomes source material for a post, clip, article, or caption.
That does not make every recording automatically good. People can still ramble, overstate, or perform. The value is that the workflow creates something inspectable before the public asset is packaged. An editor can trace a claim back to a spoken answer. A speaker can approve the implication. A clip can be checked against the surrounding context.
This is especially useful for LinkedIn because professional feeds reward signals that feel personal and credible. A post that grew out of a real recorded answer tends to carry details that are hard to fake cleanly: the exact constraint, the uncomfortable tradeoff, the narrow lesson, the exception, the phrase someone used in the room, or the reason the speaker is still unsure.
Those details are not decoration. They tell readers there is a working mind behind the claim. They also give AI a better role. AI can help prepare questions, organize the transcript, suggest excerpts, draft variations, or proofread the final text. It should not be asked to impersonate the lived expertise the post depends on.
A practical check before posting
If your team publishes expert content on LinkedIn, run a simple source check before the post goes live.
First, identify the claim. What is the reader supposed to believe after reading the post or watching the clip? Write that in one plain sentence.
Second, identify the source. Can the claim be traced to a recorded answer, transcript line, product decision, research finding, customer-approved quote, firsthand example, or cited document? If not, the post is not ready.
Third, identify the human contribution. What did the named person add that a generic model would not know: a judgment call, a caveat, a disagreement, a definition, a pattern from direct work, or a concrete example?
Fourth, identify the AI role. If AI helped, name the job internally and, when appropriate, externally: research support, question preparation, transcript organization, headline options, proofreading, or formatting. A clear role is easier to defend than a vague claim that the output is authentic.
Finally, check whether the post still sounds like something the person would stand behind in conversation. If it only works as platform-shaped copy, it may be optimized for the feed while weakening the trust that feed depends on.
Start with a real answer
LinkedIn's new feedback button is timely because it turns a soft reader reaction into a platform signal: this post feels generic, automated, or unearned. The reported feature will not settle every authorship question, and it should not be treated as a perfect detector.
But it does name the direction of travel. Professional platforms are becoming less patient with content that looks fluent but empty. They are asking for perspective, context, expertise, and real people behind the words.
REC's position is simple: let AI make expert publishing easier to prepare, organize, and reuse. Keep the person's real answers upstream of the public claim. The more crowded professional feeds become, the more valuable that source trail gets.