LinkedIn just built a snitch button for AI-generated posts, and it's built into the same platform that spent years pushing you to use AI to write those posts.
On July 30, 2026, LinkedIn quietly added a new option to the three-dot menu on every post, ad, and comment: "Seems like AI slop." Click it, and the post gets suppressed in your feed while the report feeds a signal straight into LinkedIn's classifiers, the systems that decide what gets pushed into suggested and out-of-network feeds and what gets buried.
You get a quiet thank-you message. The poster gets something less obvious: a flag in their private analytics dashboard telling them their content has been marked as likely AI slop. LinkedIn Chief Product Officer Hari Srinivasan explained the reasoning plainly, saying slop is hard to define and the definition keeps shifting, so the company wants millions of real users tuning its models rather than guessing alone.
The timing isn't subtle. In the same rollout, LinkedIn killed "enhance your post," the AI tool that rewrote your drafts into that oddly uniform, humble-brag LinkedIn voice everyone has learned to recognize and mostly ignore. Srinivasan said it's being replaced with a narrower proofreading tool, one that only fixes grammar and spelling instead of rewriting your message into something else entirely.
That's an admission, whether LinkedIn frames it that way or not.
The company spent years nudging users toward AI assistance for writing posts. Now it's asking those same users to report each other for using AI to write posts. Here's the thing: LinkedIn didn't create the writing tool and the slop problem on separate tracks. It shipped enhance your post, watched millions of people use it to produce nearly identical, algorithm-flattering paragraphs, and is now trying to detect the output of the very feature it built.
Pretty bad, according to outside research. A study called "AI in Your Feed" from Pangram Labs analyzed more than a million social posts collected between late April and the end of June 2026 and found that 41% of long-form LinkedIn posts, meaning anything over 250 words, were fully AI-generated. That's the highest rate Pangram measured on any platform it studied, and roughly double the average across the others.
Engagement paid the price. LinkedIn's own median engagement rate fell 56% year over year through 2025, a decline that lines up closely with feeds filling up with generic, AI-polished thought leadership. People stopped reading. Eventually, some of them stopped posting too.
LinkedIn also disclosed the scale of automation it's fighting separately from the slop button. Srinivasan said the platform blocks hundreds of thousands of automated spam comments every single day, and according to Fortune, it has stopped billions of other automation attempts, posting at scale, coordinated slop campaigns, in just the past couple of months.
The button has one obvious problem #
There's no reliable way to tell AI writing from a human's on sight, and LinkedIn isn't pretending otherwise. Inc flagged the flaw directly: the report button runs entirely on human judgment, which means it can flag genuinely human writing that just happens to sound polished, and it can just as easily get weaponized against posts people simply dislike. A crowdsourced slop detector is also a crowdsourced dogpile tool. LinkedIn hasn't published how many reports it takes before a post actually gets demoted, or what stops a coordinated group from mass-flagging a rival's content out of the feed entirely.
Frankly, that's the tradeoff LinkedIn is making. It would rather risk some false flags and some gamed reports than keep doing nothing while its feed fills with paragraphs that all read like the same corporate motivational poster. The company that helped normalize AI-assisted posting is now betting that its users will police each other more precisely than its classifiers manage alone. Whether that bet pays off depends on something LinkedIn can't fully control: whether people report bad writing, or just writing they don't like.
Also read: The White House Just Exempted Open-Weight AI Models From Safety Tests • Google Is in Talks to Pay Over $1.5 Billion for AI Coding Startup Mechanize • Arista Networks Crushes Earnings and Proves the AI Spending Is Real