Back to blog
How-ToSeptember 24, 2026/Sohaib G

How to Use AI for LinkedIn Posts Without Sounding Like AI

You can use AI to write LinkedIn posts without sounding like AI. The trick is to let AI handle the boring parts, like structure and first drafts, and keep the parts that make a post yours: your stories, your opinions, and your way of talking.

Most "AI-sounding" posts are not bad because a tool touched them. They are bad because the tool had nothing real to work with, so it filled the gaps with safe, generic lines.

Fix the input and the editing, and the output stops sounding like everyone else's.

Why it matters: people can tell, and they scroll past

Readers on LinkedIn have seen thousands of AI posts by now. They spot the patterns in a second:

  • Openers like "In today's fast-paced world" or "Let's dive in."
  • Words nobody says out loud, like "leverage," "unlock," "game-changer," and "delve."
  • Lists that always come in threes.
  • "It's not X. It's Y." used in every other line.
  • Every sentence the same length, like a metronome.
  • Em dashes everywhere.
  • A tidy question at the end: "What do you think? Let me know in the comments!"

When a reader sees these, they stop trusting the post. They assume nobody really thought about it, so they don't think about it either. The post gets fewer comments, and on LinkedIn, fewer comments means fewer people see it.

LinkedIn now cuts the reach of "AI slop"

This isn't just a feeling anymore. LinkedIn has confirmed it.

On July 30, 2026, LinkedIn added a "Seems like AI slop" option to the menu on every post. On the same day, it started using its own software to spot low-quality AI posts. By August 20, more than a million members had used the button.

In that update, LinkedIn's chief product officer, Hari Srinivasan, said posts LinkedIn classifies as slop now get roughly 40% fewer views. (Fortune, Moneywise)

Two things are worth knowing about that number:

  • It's about slop, not AI tools. LinkedIn is cutting the reach of posts that read as low-effort and generic. It hasn't said it punishes every post that an AI helped write.
  • One flag won't sink you. LinkedIn says no single piece of feedback decides how far a post goes. It looks at many signals together, and it has safeguards so people can't flag someone unfairly.

Readers were already pulling back before the button. An Originality.AI study of 3,368 LinkedIn posts found that posts that were likely AI-written got about 45% less engagement than posts that were likely written by a person.

So the risk is real. But the fix is the same as before: make your post sound like a real person with something real to say.

How it works: why AI writing sounds the way it does

AI models write the most likely next word. That is what makes them useful, and it is also the problem. The most likely sentence is the one you've already read a hundred times.

Two things make text feel "AI" to readers (and to AI detectors):

  1. Predictable words. The model picks safe, common phrasing. Humans pick odd, specific words because they are describing something that actually happened.
  2. Flat rhythm. Humans write a long sentence, then a short one. Then a fragment. AI tends to write every sentence around the same length.

So the fix has two parts. Give the AI specific, real material so it has something besides clichés to use. Then break up the rhythm and cut the tells when you edit.

Step 1: Start with your own raw material

This is the biggest change you can make. Don't ask AI to "write a LinkedIn post about leadership." You'll get a post about leadership that could belong to anyone.

Give it something only you have:

  • A voice memo where you ramble about something that happened this week.
  • Messy bullet notes from a client call.
  • A transcript from your podcast or YouTube video.
  • An old post of yours that did well.
  • A comment you left on someone else's post that got replies.

Talking works especially well. When you speak, you use your real words and your real rhythm. Record yourself for two minutes, transcribe it, and hand that to the AI. The draft will sound much closer to you than anything typed from a blank prompt.

Step 2: Show the AI how you write

AI can copy a style if you show it one. Paste three to five of your own posts that you wrote yourself and like. Then tell it what to copy:

  • How long your sentences usually are.
  • Whether you use contractions ("don't" vs. "do not").
  • How you open a post. A story? A blunt claim? A number?
  • Words you use a lot, and words you never use.
  • Whether you use emojis, hashtags, or line breaks between every sentence.

Write these down once as a short "voice guide" and reuse it every time. You shouldn't have to explain yourself from scratch for every post.

Step 3: Ask for structure, not a finished post

AI is great at shape. It is weaker at substance. So use it for the shape:

  • "Give me five possible hooks for this story."
  • "Pull the three strongest points out of this transcript."
  • "Put these notes in an order that builds to a point."
  • "Cut this down to under 200 words."

Then you pick. Picking is where your taste comes in. A post built from your choices sounds like you, even if a tool wrote the first version of each piece.

Step 4: Add one thing only you could say

Before you post, find the one line that proves a real person wrote this. It could be:

  • A number from your own work. ("We cut our editing time from 6 hours to 2.")
  • A name, a place, or a moment. ("On a Tuesday call with a client in Dubai...")
  • An opinion some people will disagree with.
  • A mistake you made.

Being specific is the single fastest way to sound human. AI can't invent your experiences. If the post has none of them, add one. If you can't think of one, the post might not be worth posting yet.

Step 5: Edit out the tells

Read the draft out loud. Anything you would never say to a friend, cut or rewrite. Then run through this list:

TellWhat to do
Clichéd openerStart with the actual point or the moment the story begins.
Big corporate wordsSwap for the plain word. "Use," not "leverage."
Everything in threesKeep two. Or four. Break the pattern.
"It's not X, it's Y"Just say Y.
Same-length sentencesMerge two short ones. Chop a long one. Add a fragment.
Em dashesUse a period or a comma.
Generic closing questionEnd on your point, or ask something specific.

This takes five minutes. It's the difference between a post people skim and a post people finish.

Example: before and after

Here's a typical AI draft from a blank prompt:

In today's fast-paced world, we often overthink our content. But here's the truth: it's not about perfection. It's about authenticity. The posts that perform best are raw, real, and relatable. So stop overthinking and start posting. What's the best post you almost didn't share? Let me know below! 👇

Here's the same idea, rebuilt from a two-minute voice memo:

I posted every weekday in January for 21 days.

My best one got 9,000 views and I wrote it in 5 mins

I mean it wasn't the one I spent the most time on. It was just a 60-word post about a client who paid me four months late.

This one was something I wrote on the go and I almost didn't post it

Same topic. The second one has a real story, real numbers, uneven sentences, and a clear point. No one would mistake it for a template.

Common mistakes

  • Posting the first draft. The first draft is where the clichés live. Always edit.
  • Using one-line prompts. "Write a post about X" gives you everyone's post about X.
  • Running it through a "humanizer" and calling it done. A rewrite tool can fix rhythm and word choice, but it can't add a story you never gave it.
  • Over-editing into stiffness. Don't swap every word for a fancier one. Plain and a bit messy reads as human.
  • Faking experience. If the AI invents a client story that didn't happen, delete it. Made-up details are worse than no details.

Tools that help

You can do everything above with any general AI chat tool if you save your voice guide and paste it in each time. It works. It just takes some setup every session.

Some tools build these steps in. Reshaper AI, for example, lets you save your voice once, takes a YouTube video, a transcript, or your raw thoughts as the starting point, and runs every draft through a check for common AI patterns before you see it. That saves the setup, but you still need to add your own specific details and do a final read. No tool can do that part for you.

Whatever you use, the rule is the same: real input in, real edit before posting.

Frequently asked questions

Does LinkedIn penalize AI-written posts?

It penalizes AI slop. In August 2026, LinkedIn said posts it classifies as slop get roughly 40% fewer views. That covers low-effort, generic posts, not every post an AI helped with. If your post has real ideas and sounds like you, the tool you used isn't the problem.

Is it dishonest to use AI for LinkedIn posts?

Not if the ideas and experiences are yours. Plenty of people use editors, ghostwriters, or spell check. It becomes dishonest when AI invents experiences, results, or opinions you don't actually have.

Will AI detectors flag my post?

They might flag a raw AI draft. They are much less likely to flag a post built from your own words, edited for rhythm, and full of specific details. Detectors are also not always right, so focus on how the post reads to a person first.

How long should editing take?

For a normal post, five to ten minutes. Read it out loud once, cut the tells from the table above, and add one specific detail. If you need an hour to fix it, the input was too thin. Start from better raw material next time.

Quick checklist before you post

  • Did it start from something real I said, wrote, or did?
  • Does it include at least one detail only I could know?
  • Did I read it out loud?
  • Did I cut the clichés, the threes, and the em dashes?
  • Do the sentences vary in length?
  • Does it end on a point, not a generic question?

If you can say yes to all six, post it. It will sound like you, because it is you.

Try Reshaper AI free

Turn one piece of content into posts for every platform in seconds.

Start free