"LinkedIn's 'AI slop' button is working, and the irony runs deep"

A few weeks ago LinkedIn did something unusual for a social network: it gave its members a button to tattle on bad content. The option, which sits behind every post and comment and reads "seems like AI slop," has now been clicked more than a million times since launch. In an update this week, Chief Product Officer Hari Srinivasan put a number on the effect — members are seeing roughly 40% fewer views of what LinkedIn classifies as AI slop than they did before the button existed. On its face, that's a clean win. Dig a layer deeper and it gets considerably more interesting.

Start with the scale of the problem the button is responding to. A study from AI-detection firm Pangram estimated that more than 40% of longform posts on the professional network were likely machine-written. LinkedIn became notorious for this for a reason that isn't hard to unpack: visibility on the platform is heavily weighted toward posting frequency, and generative AI collapsed the cost of posting frequently to essentially nothing. The result is a feed full of polished, interchangeable "thought leadership" that says very little. Half the content economy of a professional network had quietly gone synthetic.

The mechanics LinkedIn is using are deliberately vague — Srinivasan said the company weighs "many signals" to decide a post's reach, with "safeguards to help prevent individual feedback from being used to unfairly target other members." That's the right instinct; a crowdsourced button is trivially easy to weaponize against a rival or a critic, and the safeguard language is LinkedIn signaling it knows this. The company had already been suppressing posts that used telltale AI phrasing like "it's not X, it's Y." What's new is the feedback loop closing: users will now get a notification when one of their posts has been reported as slop, surfaced in the post-analytics panel.

Here's the first insight worth drawing out, and it's a measurement problem nobody is really talking about. "40% less views on what we classify as AI slop" is a self-referential statistic. The same classifier that decides what counts as slop is the one reporting that slop is down. There's no independent ground truth here — and the button itself is explicitly subjective ("seems like AI slop"), a feeling rather than a fact. So the honest read of that 40% figure is not "the feed is 40% cleaner" but "the posts our filter catches are getting 40% less reach." Those are different claims, and the gap between them is where the uncertainty lives.

The second insight is the irony baked into the whole exercise. This is a platform that spent the last two years actively selling AI writing to its members — most visibly through an "enhance" button that would rewrite your post or message with a single tap. LinkedIn removed that button at the same moment it rolled out the slop-reporting tool, which reads as an admission: the company was simultaneously selling the fuel and complaining about the fire. It's hard to police synthetic content with a straight face when your own product shipped a one-click generator for it. The tension isn't resolved; it's just been quietly hidden.

The third insight is about what happens next, and it's a cat-and-mouse problem. By notifying people when their posts are flagged, LinkedIn is effectively teaching its users what the filter looks for. The predictable outcome is that low-effort AI content doesn't disappear — it evolves to avoid the tells. The slop becomes less detectable, which is not the same thing as less prevalent. Every anti-slop signal a platform publishes is a specification for how to build better slop, and generative models are more than fast enough to keep pace.

Zoom out and the whole thing is a textbook tragedy of the commons. For any individual member, posting more — and using AI to post more — is rational, because volume drives reach and reach drives whatever outcome they're chasing. But the aggregate of all that rational behavior is a degraded feed that no one actually enjoys. A report button is a cheap, coarse way to push back against that dynamic without paying for human moderators at the necessary scale. The real question is whether a "seems like slop" flag is a meaningful signal or just a pressure valve — a way to make members feel heard while the underlying economics churn on.

That said, it would be a mistake to write the button off as theater. A 40% reduction in reach, even self-reported, is a real shift in how content is distributed, and a million clicks in a few weeks suggests genuine appetite for it. The Verge's write-up notes the number came from LinkedIn itself, which warrants some skepticism, but even a modest real-world effect would be notable for a change this simple. Other platforms are watching closely — every major social network has an AI-slop problem, and a cheap, scalable report button is the kind of experiment they can copy in a sprint.

The deeper story is about what LinkedIn is becoming. It has spent years drifting from "professional network" to "content platform," and that drift is precisely what made it a magnet for AI slop in the first place. When your engagement model rewards raw posting volume, and your members can summon unlimited volume from a model, you've built a machine that manufactures its own pollution. The slop button is a genuine fix for a symptom, but it doesn't touch that deeper design tension. If nearly half your longform content is synthetic, the problem isn't really the content — it's that the incentives that produced it are still in place.

None of this is doom, and it's worth keeping the right frame. LinkedIn deserves real credit for doing something concrete, for removing its own AI-enhance button, and for being transparent enough to publish the kind of numbers we can now argue about. The slop button is a rare case of a platform handing users a lever to pull instead of a blog post to read. But the most interesting question it raises isn't whether the button works — it's whether a network that monetizes posting frequency can ever truly be done with the synthetic content that frequency invites.

Further reading: Engadget's original report on LinkedIn's AI slop button · Forbes on what the slop crackdown means for your posts