BlogEssay

LinkedIn Kept the AI Editor and Dropped the Ghostwriter

You can write with AI. What can cost you the account is a tool running on linkedin.com, even when you click Comment yourself.

5 October 2026 8 min read 15 sources cited
Contents
  1. A weak post costs reach, and a tool can cost the account
  2. Slop is a post with no point, whoever wrote it
  3. LinkedIn never banned the em dash
  4. Write the point yourself, then let ChatGPT edit it
  5. A comment has to add something the post did not say
  6. Clicking Comment yourself does not make a tool allowed
  7. Myths you can stop repeating
  8. Go deeper
  9. Sources

LinkedIn now lets anyone mark your post "Seems like AI slop," and more than a million people used that button in roughly its first two weeks.16 If you draft in ChatGPT, the button is the smaller worry. LinkedIn allows writing with AI and cuts the reach of posts that say nothing, whoever wrote them.1 The rule that gets accounts restricted is about software working on LinkedIn's site, and it never asks who clicked Comment.15

In short

  • Draft and edit with AI if you like. LinkedIn judges whether the post carries a point of view from you.
  • An empty post loses reach with people outside your network. A "Seems like AI slop" flag is feedback and opens no case against you.
  • Accounts get restricted over tools. LinkedIn bans extensions that scrape, change or automate its site, even when you click the button yourself.
  • No official LinkedIn source bans the em dash or any phrase. Fix the point and most of the tells leave with the filler.

I sell for two companies and build Handful, which drafts LinkedIn comments, so I read LinkedIn's own statements to learn exactly where its lines are.

A weak post costs reach, and a tool can cost the account

LinkedIn has three separate sets of rules for bad content, each with its own trigger and its own price.

Quality ranking

Costs you reach with strangers.

Content that looks AI-generated and "lacks clear perspective" is "less likely to be widely distributed beyond a person's immediate network."9 Nothing is removed. The slop flag "does not indicate a policy violation or trigger a reporting flow."18

Automation ban

Can cost you the account.

The User Agreement bans "bots or other unauthorized automated methods" that "create, comment on, like, share, or re-share posts."48 Members who use banned tools "risk having their accounts restricted or shut down."14 A brilliant comment posted by a script is still a scripted comment.

Spam rules

Catch people with no software at all.

"Excessive, irrelevant, or repetitive comments or messages" count as spam.13 Paste one friendly template under fifty posts in an afternoon and you can trip it by hand. Pods count too: members should not "agree with others ahead of time to like or re-share each other's content."7

Most bad advice comes from mixing these up. People grow afraid of a proofreader, which only ever meets the first rule, then trust a reply extension because its comments read well, though the second rule never looks at quality.

Save this

Six checks before you post, comment or install

  1. Say your point in one sentence a colleague could disagree with.
  2. Add one thing only you have: a number, a decision, a customer's exact words.
  3. Let AI tighten your words. Never let it invent a story, a result or an opinion.
  4. If your comment would fit under anyone else's post, rewrite it.
  5. Does a tool read, change or act on linkedin.com? Remove it, even if you click Comment yourself.
  6. Skip pods, comment swaps and fifty comments in an afternoon.

handful.cc/blog/linkedin-ai-slop

Slop is a post with no point, whoever wrote it

LinkedIn's Help Center puts its position in one line: "LinkedIn's focus is not on how content is created, but whether it adds value." Slop, in its definition, "may sound polished on the surface but lacks a clear point of view, unique perspective, or substance."1 Polish is no defense, and a person can write slop without ever opening a chatbot.

“It's ok to use AI to help you write, but your posts and comments need to represent your voice and your perspectives. The ultimate value comes from the human behind the tool.”9

Laura Lorenzetti, VP and Executive Editor, LinkedIn

In July LinkedIn said it was pulling the "enhance your post" rewriter from its own composer for a feature "that proofreads your words, but does not change your voice."10 LinkedIn kept the editor and dropped the ghostwriter. Do the same with your drafts.

LinkedIn never banned the em dash

You have seen the lists: delete em dashes, one-line dramatic paragraphs and "it's not X, it's Y." I looked for each in LinkedIn's own statements and found none. The closest LinkedIn has come is sameness, too many posts and comments "written in the same AI style," as its chief product officer, Hari Srinivasan, put it.53

Readers may still notice the tells. A post with a real point sheds most of them while you write it, and a post without one stays empty however carefully you scrub it.

Write the point yourself, then let ChatGPT edit it

A chatbot in another tab never touches LinkedIn, so the automation rules ignore it.15 The risk sits on the quality side, and it comes from the order you work in.

Ask a model for "a LinkedIn post about discovery calls" and you get something like the first draft below, fluent and sensible and empty. The model has never run a discovery call. You have.

Before (an illustration I wrote)

Discovery calls are where deals are won or lost. Too many reps rush to the demo instead of truly understanding the buyer's pain. The best sellers slow down, ask better questions and let the prospect talk. Curiosity builds trust, and trust closes deals. What's one question you always ask on discovery calls?

After (an illustration; the clinic is invented)

We sell scheduling software to clinics. For a year our reps opened discovery with "What are your biggest challenges?" and heard the same three answers. In March we started asking how many appointments the front desk rebooked by phone last week. Most office managers didn't know. The ones who checked came back with the number we built the proposal around. Which question gets your buyers to go look something up?

Any sales trainer could sign the first draft. Only someone who changed one question and watched what happened could write the second.

So reverse the order. Write the raw material first, badly if you have to: the point you would defend, what happened, the number, the part you are unsure about. Then give the model editing work.

Check every name and number the model touched. If you leaned on AI heavily and that isn't obvious from context, LinkedIn recommends saying so, though it does not require it.1

A comment has to add something the post did not say

Comments are where most sellers should spend their LinkedIn time, and the bar is higher than for posts, because the author reads yours closely.

LinkedIn's Rishi Jobanputra describes a low-quality comment as one that repeats "what's already in the post without adding new ideas."17 "Great insights!" is the easy case. The harder one is a well-written paragraph that summarizes the post back to its author, which is what a model tends to produce when you paste in a post and ask for a comment.

Say a VP of Sales posts that her team stopped cold calling this year and pipeline went up.

Before (an illustration I wrote)

This is such an important point. Dropping cold calls freed your team to focus on warmer channels, and the pipeline numbers show it paid off. More sales leaders should question the activities they've always done.

After (an illustration I wrote)

Same for us, but only with companies under about 200 people. Above that, calls still booked a third of our first meetings, mostly because whoever picked up could route us to the right person. Did you split your results by company size?

The polite version hands her own post back to her. The other gives her something to go and check.

What to add when you have no data of your own is in the piece on comments.

Clicking Comment yourself does not make a tool allowed

In February Gyanda Sachdeva, LinkedIn's VP of Product Management, defined automated comments as ones posted "without human action involved in clicking the 'comment' button."11 Many people read that as a permission slip. A second rule in the Help Center bans "third-party software or browser extensions that scrape, modify the appearance of, or automate activity on LinkedIn's website,"15 and it never asks who clicks.

So a tool has to pass two tests. A person presses the button, and the tool stays off LinkedIn's pages. An extension that reads your feed to suggest replies, or puts a "draft reply" button under each post, fails the second. LinkedIn has not ruled on that exact case, so this is my reading, and if one is in your browser I would remove it today. The piece on AI tools sorts the rest.

Handful never logs in or touches your account. It shows up to five posts a day, fewer on a thin day, and you comment by hand.

The whole piece fits in one line. AI may edit you, it may not be you, and no tool of yours should touch linkedin.com.

Myths you can stop repeating

The piece on the slop numbers reads each headline figure word by word.

Myth

LinkedIn detects AI writing with 94% accuracy.

In initial testing LinkedIn was "correctly identifying generic content 94% of the time." That is generic content, with no false-positive rate.9

Myth

A slop flag cuts your post's reach by 40%.

The 40% is a platform-wide drop in views of what LinkedIn classifies as slop, compared with a few weeks earlier. LinkedIn published no per-post penalty.3

Myth

360Brew is LinkedIn's 2026 algorithm.

Asked if it is part of the ranking system, LinkedIn's VP of Engineering said "Short answer: No!" It was a small 2025 test, since shut down.54 What the live feed reads is in the piece on feed ranking.

Go deeper

Sources

  1. 01LinkedInBest practices for content created with the help of AI, LinkedIn Help Center, ~2026-08 (live by 2026-09-13). linkedin.com
  2. 03LinkedInFollow-up on "Seems like AI slop" ("One more follow up on AI Slop"; duplicate of [19]), Hari Srinivasan, CPO, LinkedIn post, 2026-08-19. linkedin.com
  3. 07LinkedInProfessional Community Policies, LinkedIn, undated (© 2026). linkedin.com
  4. 09LinkedInKeeping conversations real on LinkedIn, Laura Lorenzetti, LinkedIn Pulse, 2026-05-20. linkedin.com
  5. 10LinkedInAI slop is a top priority for all of us, Hari Srinivasan, CPO, LinkedIn post, 2026-07-30. linkedin.com
  6. 11LinkedInFriday update: automated comments, Gyanda Sachdeva, VP Product, LinkedIn post, 2026-02-13. linkedin.com
  7. 13LinkedInSpam, LinkedIn Help Center, ~2024. linkedin.com
  8. 14LinkedInProhibited software and extensions, LinkedIn Help Center, ~2024. linkedin.com
  9. 15LinkedInAutomated activity on LinkedIn, LinkedIn Help Center, ~2026-10-02. linkedin.com
  10. 16LinkedInHow LinkedIn is Continuing to Tackle AI Slop, LinkedIn Pressroom, 2026-09-15. news.linkedin.com
  11. 17LinkedInMaking LinkedIn conversations more relevant and valuable, Rishi Jobanputra, LinkedIn Pulse, 2026-07-31. linkedin.com
  12. 18LinkedInShare feedback on AI slop, LinkedIn Help Center, ~2026-09. linkedin.com
  13. 48LinkedInUser Agreement (§3.6 Generative AI, §8.2 Don'ts), LinkedIn, 2025-11-03. linkedin.com
  14. 53LinkedInOn AI slop... First, I see it too, Hari Srinivasan, CPO, LinkedIn post, 2026-05-21. linkedin.com
  15. 54LinkedIn360Brew clarification, Tim Jurka, VP Engineering, LinkedIn post, 2026-04-14. linkedin.com

Comments that add something, on the posts that matter.

Handful finds up to five LinkedIn posts a day from people who need what you sell, and drafts a comment in your voice. You read it, edit it, and post it yourself.

Start free →