LinkedIn Says Fake Activity Jumped 46%. If AI Writes Your Posts, Read This.

A guy at a networking breakfast in Perrysburg told me a couple weeks ago that his LinkedIn had flatlined. Same posting schedule he'd kept for two years, same kind of content, and sometime in the spring his numbers fell off a cliff. He wanted to know if he'd been shadowbanned.

He hadn't. He started having ChatGPT write his posts in February.

LinkedIn just put numbers behind what he was feeling. In its latest reporting they said they caught 46% more inauthentic activity in the first half of this year than in the six months before it. Fake profiles, automated engagement, accounts that exist only to like things, and posts that were written by a bot and dropped on the feed without a human reading them first. That's what they already found and took down.

What LinkedIn is actually catching

This is where people get it wrong. There's no magic detector sitting there going "written by AI, delete it." What the ranking system watches is whether real people stop and read the thing. Did anybody get past the first two lines. Did anybody leave a comment longer than one word.

Generic AI writing fails that test almost every time, because it reads exactly like the other four hundred posts that scrolled by that morning. Same rhythm, same little dash stuck in the middle of every sentence. Nobody stops on it, so LinkedIn quietly stops showing it. The numbers people are throwing around put the hit somewhere near half the reach a normal post would get.

So the tool that was saving you twenty minutes in the morning is quietly taking away the thing you were posting for.

Why this hits harder here

Most of the business owners I work with around Toledo aren't on LinkedIn to build some personal brand. They're on there because that's where their referral network sits. The commercial insurance guy, the shop owner, the CPA, the HVAC company chasing property management contracts, the lawyer who gets half his work from other lawyers. Their whole value on that platform is that a few hundred local people see them show up regularly and remember who they are.

When your posts stop reaching those few hundred people, nothing dramatic happens. The referrals just go quiet, and you don't notice for two or three months, and by then you're blaming the economy.

What I'd actually do this week

Pull up your last ten LinkedIn posts and read them out loud. If you can't tell which ones you wrote and which ones the AI wrote, you found your problem.

Then change how you're using it. I still run almost everything I write through AI. The difference is I don't ask it for the post. I dump my actual thought in first, messy, half a paragraph, whatever I'd say to somebody sitting in my truck, and then I ask it to fix my typos and tighten it up. What comes back still sounds like me because the thinking was mine. The AI didn't have the opinion. I did.

The other thing that works is specifics. A real number, a real customer situation, a street name. AI can't invent those for you, and that's exactly why they land. "Five ways to improve your customer service" gets buried. "We changed how we handle after hours calls and it added eleven jobs in August" doesn't.

And watch the fingerprints. If your posts are loaded with em dashes and you've never typed one in your life, people notice, even if they can't name what's off. Same with words like delve, and any sentence built like "it's not about X, it's about Y." I hand out a list of these at my workshops because once you can spot them you can't stop spotting them, and you start seeing them in your own drafts.

You can keep using AI on LinkedIn. Just stop letting it do the part that was supposed to be you.