LinkedIn is coming for your AI posts. Good.

LinkedIn's content volume is up 14% year over year. Laura Lorenzetti, VP and Executive Editor at LinkedIn, said so herself this week — and then explained why that's a problem, not a win.

The platform announced a new AI detection system built to identify and suppress what it's calling "AI slop": low-effort, AI-generated posts that sound polished but say nothing. In early tests, the system flagged generic AI content correctly 94% of the time. Posts don't get removed. They get buried — often not visible beyond a poster's first-degree connections.

If your B2B content workflow pipes a topic into Claude, cleans up the output in five seconds, and schedules it to post on Tuesday morning, you're exactly who this targets.

What LinkedIn is actually detecting

The detection isn't a simple "is this AI?" binary. LinkedIn's system, trained in collaboration with its in-house editorial team, is looking for pattern signals: templated phrasing, predictable structures, zero personal anecdotes, the same unnaturally polished grammar that AI defaults to across millions of accounts.

It's also going after comments. Bot-generated replies that parrot the original post. Generic "Great insight!" engagement at scale. Automation tools cranking out responses with no human input. Those get suppressed too.

The line Lorenzetti drew is meaningful: AI as a writing aid is still fine. The system isn't punishing you for using Claude to sharpen a sentence or restructure a draft. It's punishing you for letting the model do all the thinking.

Here's the irony that The Decoder flagged and I can't stop thinking about: Microsoft — LinkedIn's parent company — simultaneously rolled out a new Copilot feature in the browser that actively promotes AI-assisted writing, with LinkedIn as the demo platform. The demo post was, reportedly, maximally generic. The left hand and right hand aren't talking to each other. That tells you something about how fast this is all moving.

The take everyone's missing

Most of the coverage treats this as a reach problem. You post AI slop, you lose impressions. Fine. But there's a second consequence that matters more for B2B marketers, and I haven't seen anyone connect it directly yet.

LinkedIn is currently the #1 citation source across every major AI search engine — above Wikipedia, above every news publisher. Semrush tracked 325,000 prompts across ChatGPT, Perplexity, Gemini, Google AI Mode, and Copilot earlier this year. LinkedIn came out on top everywhere. 14.3% of ChatGPT responses cite it.

That's the channel your buyers are using to shortlist vendors. G2's recent report on AI search found that many B2B buyers are now "one-shotting" shortlists with a single chatbot prompt — comparing vendors, summarising strengths, returning a usable list before they've visited a single website.

If your LinkedIn presence is suppressed generic content, it's not just your feed reach that suffers. You're falling out of the citation pool that shapes the first impression a buyer gets when they ask ChatGPT who to shortlist.

What actually survives the filter

LinkedIn's 360Brew algorithm is specifically looking for the opposite of AI defaults: dwell time, saves, substantive comments that add to the conversation. The system evaluates lexical diversity — human writers naturally vary rhythm and structure. It checks profile-content alignment, whether what you post matches your stated expertise. It rewards specificity and original perspective.

The posts that survive aren't necessarily short. They're posts that contain something the model couldn't have generated without you in the room. A specific number from your own data. A take that contradicts the obvious read. A mistake you made and what happened next.

I run marketing ops at a Copenhagen SaaS. When I post about something I actually built in n8n or ran into inside HubSpot, the engagement is structurally different from anything I've ever scheduled from a polished template. Not because of the topic — because the model of that post is me, not a generic professional.

The rollout is gradual. Lorenzetti said it could take several months before users see the full impact. But some creators are already reporting a 50% year-over-year organic reach drop, while people posting authentic, expert-level content are seeing stronger results than before. The gap is widening, not closing.

LinkedIn isn't banning AI. It's ending the free ride for people who outsourced their perspective to a language model and called it a content strategy.

The question worth sitting with: how much of your current LinkedIn content would survive a 94% accurate "does this actually sound like you?" filter?