When a million people voluntarily click a button that says "this feels fake," you're watching a platform admit it lost control of its feed.
The Summary
- LinkedIn launched a "seems like AI slop" button on July 30, and over 1 million users have already clicked it to flag suspected AI-generated content
- Views on suspected AI slop posts are down 40% compared to a few weeks ago, according to LinkedIn's chief product officer
- The move came after AI detector Pangram flagged 41% of LinkedIn's longform posts as fully AI-generated
- LinkedIn is betting on human judgment over automated AI detection, with flagged users now receiving direct feedback
The Signal
LinkedIn just ran the numbers on its own feed and didn't like what it found. When an external AI detector says four out of ten longform posts on your platform are machine-generated, you've got a content authenticity crisis. So the company did something unusual: it asked users to become the detection layer.
The "seems like AI slop" button lives in the three-dot menu on every post. Click it, and you're telling LinkedIn this content feels synthetic. No algorithm, no detector model. Just human intuition about what sounds real and what sounds like a ChatGPT fever dream about "synergizing cross-functional stakeholder paradigms."
"We want members to get feedback from real humans on what sounds authentic, not just have an AI detector review it and get it wrong."
The results are striking. Over a million users clicked the button in less than a month. That's not passive scrolling. That's active disgust at scale. And LinkedIn's chief product officer Hari Srinivasan says views on flagged content dropped 40% in the weeks since launch.
Here's what makes this different from every other platform's AI moderation playbook:
- LinkedIn is surfacing the flags to the original posters, not just burying content silently
- The company explicitly chose human feedback over automated detection
- They're treating AI slop like a social norm violation, not a terms-of-service issue
Srinivasan acknowledged there's more work to do to "ensure LinkedIn remains a place where you can find real people and real perspectives." That phrasing matters. LinkedIn isn't fighting spam or bots. It's fighting the erosion of authenticity on a platform built around professional identity.
The timing is revealing. LinkedIn rolled this out in late July, right as generative AI tools became sophisticated enough that slop isn't obviously slop anymore. Early AI-generated posts were easy to spot. The new stuff sounds plausible until you realize every paragraph says nothing in 47 words. LinkedIn is training its users to notice the uncanny valley of business prose.
The Implication
Every social platform will face this choice: automate the detection or enlist the humans. LinkedIn picked humans, which means they're betting that authenticity is a community standard, not a technical problem. If you're using AI to draft posts, you now have a direct feedback loop telling you when you sound like a bot. That's either a forcing function for better prompts or a signal to stop automating your voice entirely.
Watch how other platforms respond. If LinkedIn's 40% drop in slop views holds, expect Twitter, Meta, and others to test similar crowd-sourced flagging. The "seems like AI" button might become as common as the like button. Which means the next frontier isn't whether AI can write like a human. It's whether humans can tell the difference at scale.