TL;DR:
Ever scroll past a LinkedIn post that just feels off: the one-line paragraphs, the bulleted "key takeaways," the emoji doing a lot of heavy lifting? You're not imagining it, and now LinkedIn is giving people a way to say so out loud.
LinkedIn is testing a report option that says, plainly, "Seems like AI slop."
404 Media discovered it on July 30 by clicking the three-dot menu on a post, tucked in alongside the platform's other reporting categories. Click it, and the post disappears from that user's feed, along with a quick thank-you message.
LinkedIn hadn't responded to a request for comment on the specific feature at the time of the report, so treat this as an unannounced, possibly limited rollout, not a confirmed platform-wide launch.
It's a reporting option, not a detection algorithm you can see working. When a post has the tells (even spacing between one-sentence paragraphs, a bulleted "key takeaways" list, heavy emoji use, the classic "It's not X, it's Y" construction) users can flag it the same way they'd report spam or harassment.
LinkedIn's response is immediate: the post disappears from that user's feed. What happens on the back end after multiple reports (downranking, account review, or nothing beyond that one feed) hasn't been disclosed.
A lot, according to new research. AI detection company Pangram Labs ran a Chrome extension that passively scanned more than a million posts people actually scrolled past on LinkedIn, X, Reddit, Substack, and Medium over roughly two and a half months in 2026.
The findings, published on Pangram's blog, were stark:
Worth a caveat: that "highest rate" applies specifically to fully AI-generated writing. Count AI-assisted or mixed content too, and X/Twitter's combined total was actually higher, with only 53% of its long articles registering as fully human. LinkedIn's problem isn't the only one out there; it's just the most visible.
LinkedIn accounted for 62% of all the AI-generated content Pangram flagged, across five platforms combined, despite making up only about a third of the posts scanned.
Pangram's CEO, Max Spero, told 404 Media the data is likely a floor, not a ceiling: people who install an AI detector in the first place are probably already trying to avoid AI content.
No, and that distinction matters. LinkedIn still builds AI writing tools directly into the platform: its assistant, now branded "Enhance post," still helps people draft and edit. LinkedIn isn't banning AI-assisted writing; it's targeting a narrower category.
A LinkedIn spokesperson told 404 Media the company is working to "reduce low quality, automated or generic content" while supporting AI as a way to "beat the blank page problem."
Translation: using AI to help you write is still fine. Posting something that reads generic, low-effort, or mass-produced is what's getting flagged, whether or not AI actually wrote it.
Because two months of promising to fix this hadn't visibly worked. In May 2026, LinkedIn announced a policy shift called "Keeping Conversations Real," pulling its AI writing assistant out of the main compose button and pledging to downrank generic, automated-sounding content.
Pangram flagged an ironic detail: the executive's announcement post about cracking down on AI slop was itself detected as AI-generated. Two months later, with the research showing the problem was still severe, LinkedIn added a second lever: putting the judgment call directly in front of users instead of waiting on its own detection systems to catch everything.
Reddit took a different approach. Rather than adding a report button, it leaned into being provably human: publishing its own post about keeping the platform "real and safe" and launching an ad campaign built around the idea that people, not bots, make the site work. Two platforms, one problem, two fixes: LinkedIn crowdsourcing the judgment call, Reddit selling humanness as the product itself.
Here's the part that actually matters: The headline fact isn't that LinkedIn has an AI problem. It's that LinkedIn's own users can now call it out, post by post, in public, whether the account posting is a solo creator or whoever runs a sports org's or a small business's LinkedIn page.
Skepticism toward written content isn't a fringe reaction anymore; it's a built-in feed feature. For anyone whose LinkedIn posts double as a portfolio or a pitch, that's not just a platform-trust problem, it's a "why would this prospect keep reading you" problem.
Live video sits on the other side of that shift. You can't fake a live, unscripted stream the way you can fake a paragraph of thought leadership. A multicam livestream, a real-time Q&A, or a behind-the-scenes broadcast is dramatically harder to fake, and an audience can test it in real time in the comments in a way they can't test written text.
For streamers already building an audience on live formats, that's a tailwind. The platforms creators write on are working to make static, AI-polished text less trustworthy, while live video keeps its built-in authenticity advantage.
That doesn't mean AI tools are off-limits. Using AI to draft a caption or clean up a title is nothing like letting it write your on-camera voice for you. The behavior LinkedIn just built a button around ("I don't trust this because it doesn't sound human") shows up everywhere content competes for attention, not just on LinkedIn. If this button changes how creators think about proving they're real, live video is about to matter a lot more than it already does.
No. LinkedIn still offers a built-in AI writing tool ("Enhance post") and hasn't banned AI-assisted writing. It's targeting content that reads as generic or automated, not AI use itself.
LinkedIn hasn't detailed its detection method. Its May 2026 policy said it would start downranking generic content algorithmically, and the new "Seems like AI slop" option adds user-flagging on top of that.
A single reporting button won't fix LinkedIn's AI slop problem, but it's a signal worth paying attention to: platforms are starting to build tools that let audiences act on distrust, not just tolerate it.
For anyone posting to build an audience or a reputation, the safer bet isn't avoiding AI tools altogether, it's leaning into formats that carry their own proof of authenticity. Live video does that by default.
An audience watching something happen in real time doesn't need a button to decide it's real.
That's the problem Switcher is built around: making multicamera livestreams, real-time Q&As, and behind-the-scenes broadcasts easy to produce without a full crew. If this shift has you thinking about adding more live content to your own mix, it's worth a look.