AI Text Cleanup for LinkedIn: The 5-Point Checklist Before You Post

A checklist for turning a ChatGPT or Claude draft into a LinkedIn post that reads as yours: em dashes, emoji spam, bold markdown, hidden characters, breaks.

A LinkedIn post drafted by ChatGPT, Claude or Gemini needs about two minutes of cleanup before it is fit to publish. Not because the writing is bad, but because the draft was formatted for a chat window and LinkedIn is plain text, and because a few habits of machine writing are now widely recognised. This page is the checklist. Work through the five points in order and the post will read as something you wrote.

1. Em dashes

Count them. An AI draft of 200 words will often contain four or five em dashes, the long dash that joins two clauses with no spaces. Readers of LinkedIn have been told for a year that this is the sign of a generated post, and comments calling it out are common. Keep at most one, where it does real work, and turn the rest into a comma, a full stop or a colon depending on the sentence.

The em dash guide covers find-and-replace and the reasons behind the habit. For a post-sized piece of text the quickest route is to paste it into the AI Text Cleaner, which replaces each dash with a comma by default and can split the sentence instead if you prefer shorter lines.

2. Emoji spam

AI drafts use emoji as bullets: a rocket for growth, a checkmark for each point, a light bulb before the lesson. On LinkedIn this pattern is so familiar that it reads as a template. It also renders differently on every device, and a screen reader reads each one out by name, so a post with fifteen of them is hard work for anyone using one.

Remove all of them, then add back one or two if the post needs a visual break. Never use them as list markers. The emoji remover can strip them all, or replace each with a placeholder so you can decide one by one, and the full cleaner removes them as part of its default pass.

3. Bold markdown and other symbols

LinkedIn does not render markdown. Every **, every #, every backtick and every set of link brackets appears in the published post exactly as typed. This is the most visible mistake because it looks broken, not just generated.

Strip the markdown and keep the words. Headings become a short line on their own. Bold lead-ins ("Lesson 1:") become plain text, which works fine because the colon already does the job. Links go in as bare URLs, which LinkedIn turns into a preview card for the first one. The cleaner's markdown option does all of this and, if you tick keep list markers, leaves your hyphens so the list still scans as a list.

Resist the temptation to replace bold markdown with the Unicode bold letters from a text-generator site. They look bold but they are separate characters that search cannot match and assistive technology cannot read properly.

4. Hidden characters

Text copied from a chat window can carry zero-width spaces, non-breaking spaces and other invisible characters. On LinkedIn they cause two specific problems. A zero-width space inside a hashtag or an @mention stops it from linking, so #hiring looks right but is not clickable. A non-breaking space is treated as a normal character rather than a break, which can put the See more fold in the middle of a word.

You cannot see them, so check rather than trust. The hidden character checker highlights each one with its code point. The cleaner removes them in the same pass as everything else, and the counter under the output tells you how many it found.

5. LinkedIn line-break style

LinkedIn posts are read on phones, in a narrow column, with the first two or three lines visible before the See more fold. The formatting that works is short paragraphs of one to three sentences, a single blank line between them, and a first line that stands alone. Chat drafts tend to come out either as long paragraphs, or as bullet lists with no spacing, or with runs of blank lines that LinkedIn collapses anyway.

Two cleanup rules cover most cases. First, keep exactly one blank line between paragraphs; the cleaner's whitespace option reduces bigger gaps to one and leaves single ones alone. Second, if the draft is a bullet list, consider turning it into short paragraphs instead. A list of eight one-line points reads as a slide, while eight short paragraphs read as a person talking. If the text has no line breaks at all, or line breaks in the wrong places from a PDF or email, the line break tool can rebuild the paragraphs.

Check the first line last. It should be a complete thought that makes sense on its own, because for most readers it is all they see before deciding whether to expand the post.

The two-minute routine

1. Copy the draft from the chat window.
2. Paste it into the AI Text Cleaner with the default options. Tick keep list markers if you want to keep a list.
3. Read the "changes made" line. It reports how many dashes, markdown markers, emojis and hidden characters were removed, which is a quick sense of how much the draft needed.
4. Read the output once. Fix the two or three sentences where a dash was doing real work, add back an emoji if you want one, and tighten the first line.
5. Copy, paste into LinkedIn, and check the paragraph breaks in the preview before you post.

What not to clean

Do not strip the personality along with the symbols. The checklist removes formatting and tics, not the point you were making. If the draft has a specific example, a number or a line of dialogue, those stay. Curly quotes and apostrophes are fine on LinkedIn; the cleaner straightens them by default because that is safer for code and data, and either style looks normal in a post. Straight quotes only matter when the text is headed for a spreadsheet or a code editor, which the smart quotes guide explains.

Last updated 2026-09-16. Everything on this page runs in your browser. Nothing you type or upload leaves your device.