LinkedIn's AI slop report button passes 1 million clicks

LinkedIn first announced the button on July 30. A few weeks earlier, AI detection service Pangram had published a blog post claiming that 41% of long-form posts on LinkedIn were fully AI-written, a finding that got wide pickup via 404 Media. If four out of every ten posts disguised as job announcements or career advice were barely touched by human hands, it's easy to see why users started losing trust in their feeds altogether.

Alongside the button, LinkedIn made two other changes. It swapped in a "new and improved" classifier for detecting AI-generated posts, and it scrapped the "polish my post with AI" feature entirely. In other words, right as it needed to filter out AI-written content, the company pulled the plug on a tool that helped users generate AI-written content in the first place. When Srinivasan first announced the button, he said, "AI slop is a top priority for all of us" — and based on this latest update, that wasn't just talk.

How to use it

Where to find it — There's no separate settings menu to dig through. The option lives inside the three-dot (⋯) menu at the top right of any post in your feed.

Step-by-step

1. While scrolling your feed, spot a post with promotional-sounding language or repetitive, awkward phrasing.
2. Click the three-dot menu at the top right of that post.
3. Select "Seems like AI slop" from the list.
4. Once reported, it feeds into LinkedIn's internal tracking system.

Who can use it — Nothing in the original announcement mentions country or plan restrictions; it's described simply as an option available to anyone from the post menu.

What you might use it for — Think of a "self-improvement quote" post that reuses the same sentence structure over and over, or a job-review post that reads like a list with no real personal experience behind it. Since accumulated reports reduce a post's visibility, flagging one is a direct action that shapes feed quality.

Views dropped as reports piled up

Srinivasan also revealed that views on posts LinkedIn classified as AI slop have dropped 40% compared to a few weeks ago. With the report button and the new classifier working together, exposure for content that looks mass-produced by AI has genuinely declined.

Date	Change
Early 2026	LinkedIn signals plans to crack down on mass-produced, barely human-edited comments
2026-07-30	Launches "Seems like AI slop" report button, rolls out new classifier, removes "polish with AI" feature
2026-08-20	Announces 1 million+ reports, 40% drop in AI slop views, previews author notification feature

Next up: notifying authors

LinkedIn isn't stopping there. It's now working on a feature that would message post authors directly, showing a note like "Some members feel this post seems AI-generated." Srinivasan explained the intent: "We're approaching this assuming good faith, and honestly, I've become more conscious myself about not sounding like AI when I write. The goal is to give people useful feedback." The tone isn't punitive toward people who get flagged — it's meant to prompt a second look at writing habits.

Editor's take

What makes this button interesting is that LinkedIn solved the problem through crowd reporting rather than detection. Classifier accuracy for AI-generated text remains a genuinely contested issue, and instead of fighting that accuracy battle, LinkedIn chose to treat the judgment of a million users as its data signal. When Pangram put out that 41% figure, LinkedIn essentially had two paths: refine its own detection model further, or absorb human intuition directly as a signal. Choosing the latter is, in a way, an admission that text-detection technology still isn't good enough on its own.

Around the same time, X faced similar criticism over AI spam, which suggests this isn't a LinkedIn-only problem but a challenge facing text-based social platforms broadly. Still, the two platforms' response speeds look quite different. LinkedIn shipped the button within weeks of the research coming out, and within a month it had published both a usage figure — 1 million clicks — and an effectiveness figure — a 40% drop. Releasing user-response data this quickly reads as a direct message to the market: we're not sitting on this problem.

For teams running content platforms or communities domestically, the lesson here isn't really about classifier accuracy — it's about where the report button lives and what it's called. Simply placing an intuitively worded option inside the three-dot menu was enough to generate a million data points, which suggests that fixing up a user-reporting channel can be more cost-effective than building an elaborate AI-detection system from scratch. For smaller platforms without much content-moderation staff, it makes sense to try low-cost, high-impact tools like this first.

In the coming weeks, once the author-notification feature actually goes live, some pushback or complaints from flagged users seem likely. And if reports start getting misused — say, someone flagging a post as "AI-like" simply because they disagree with its opinion — LinkedIn will probably have to revisit the system all over again.