UnmarkPro
AI Text Guide

Why Does ChatGPT Use So Many Em Dashes?

The long dash shows up in almost every ChatGPT answer. Here is what is known about why, the setting that helps, and how to clean the dashes and stray asterisks out of text you already have.

Short answer: Nobody outside OpenAI knows exactly why. You can ask ChatGPT to stop in its custom instructions, and for text you already have, paste it into the ChatGPT em dash remover and copy the clean version. It is free, needs no account, and your text stays on your device.

What an em dash is

An em dash is the long dash: —. Writers use it to add a pause, an aside, or a sudden turn in a sentence. It is longer than a hyphen (-), which joins words, and longer than the en dash (–), which is used in ranges like 1–5.

There is nothing wrong with the em dash. Good writers have used it for a long time. The problem is how often ChatGPT uses it. Two or three in a short message makes the text feel like it came from a chatbot, and many people now notice that.

Why ChatGPT uses so many

Here is the honest part: nobody outside OpenAI knows for sure, and OpenAI has not given a full explanation. People who study these models point to a few likely reasons:

It is probably a mix of these. The useful part is what you can do about it.

Fix 1: Tell ChatGPT to stop

For a long time, asking ChatGPT not to use em dashes did not work well. In November 2025, OpenAI's CEO Sam Altman posted that if you tell ChatGPT not to use em dashes in your custom instructions, it now does what you ask. To set it up:

  1. Open ChatGPT and go to Settings.
  2. Open Personalization and find Custom instructions.
  3. Add a line such as: Do not use em dashes. Use commas or periods instead.
  4. Save, then start a new chat.

Putting it in custom instructions works better than asking in the middle of a long chat. Still, check the result. A dash can slip through, and this does nothing for text you already copied. Menu names in ChatGPT can move between updates, so if you can't find the option, search the settings for "custom instructions".

Fix 2: Remove the dashes in one paste

For text you already have, this is the fastest way:

  1. Copy the answer from ChatGPT.
  2. Open the ChatGPT em dash remover and paste it in.
  3. Choose what each em dash should become: a comma, a spaced hyphen ( - ), a period, or nothing.
  4. Turn on Show changes if you want to see each edit.
  5. Press Copy or download a .txt file.

Number ranges such as 1–5 or Mon–Fri keep their short dash, and text inside code blocks is left alone. Your words stay the same and in the same order. Only punctuation, formatting symbols and spacing change.

A comma works for most dashes. Read the result once, though. Some sentences read better split into two, so a period can be the better pick.

The asterisks problem in Word, Docs and LinkedIn

The em dash has a cousin: stray asterisks and hash marks. ChatGPT writes in markdown, a simple way to mark bold text and headings. In the chat window, **this** shows as bold. When you paste it into Word, Google Docs, Outlook or a LinkedIn post, you often see the raw symbols instead.

The same tool fixes this. Leave Remove markdown on, and bold markers, headings and code fences turn into plain text, while bullet points become simple lines. It can also straighten curly quotes, tidy extra blank lines and remove hidden characters that make pasted text act oddly. For more on those hidden characters, read how to remove hidden characters from AI text.

Does this hide that you used AI?

No, and it is not meant to. Many people now treat the em dash as a sign of AI writing, but detectors look at much more than punctuation, such as word choice and sentence patterns. Changing dashes will not reliably change a detector score. We explain the limits of those checks in why AI detectors are mostly guesswork.

Use the cleaner so your text pastes cleanly and reads the way you like. If your school, job or platform asks you to say when AI helped, keep doing that.

What this cannot do

  • It does not rewrite or reword your sentences. It only fixes punctuation, formatting and spacing.
  • It does not make text "undetectable" or change how an AI detector scores it.
  • It cleans up to 250,000 characters at a time. Split longer documents.

Frequently Asked Questions

Why does ChatGPT use so many em dashes?

Nobody outside OpenAI knows for sure. The most common explanations are that the dash is frequent in the books and articles these models learn from, that it fits almost anywhere in a sentence, and that people who rated answers during training may have liked writing that used it. These are educated guesses, not confirmed facts.

Can I tell ChatGPT to stop using em dashes?

Yes. In November 2025, OpenAI's CEO Sam Altman said that if you tell ChatGPT not to use em dashes in your custom instructions, it now follows that request. Add a line such as "Do not use em dashes" in ChatGPT's personalization settings. It helps a lot, but a dash can still slip through, so check before you paste.

How do I remove em dashes from text I already have?

Paste the text into the ChatGPT em dash remover on UnmarkPro and pick what each dash should become: a comma, a spaced hyphen, a period, or nothing. Then press Copy. Number ranges like 1–5 keep their short dash.

Why do asterisks show up when I paste ChatGPT text into Word or LinkedIn?

ChatGPT writes in markdown, a simple formatting style. Its chat window turns **text** into bold, but Word, Google Docs, email and LinkedIn often show the raw symbols. The em dash remover can turn markdown into plain text at the same time it fixes the dashes.

Will removing em dashes make my text undetectable as AI?

No. AI detectors look at word choice and sentence patterns, not only punctuation. Removing dashes or asterisks will not reliably change a detector score. The tool is for clean pasting and the look you want, not for hiding AI use.

Is my text uploaded anywhere?

No. The text stays on your device. The tool is free and needs no account. It cleans up to 250,000 characters at a time, so split very long documents into parts.

Sources: Sam Altman's post on X (14 November 2025), as reported by TechCrunch and Tom's Guide. The reasons for ChatGPT's em dash habit are not confirmed by OpenAI; the explanations above are common theories as of October 2026. Questions: team.kyiron@gmail.com

Related Guides & Topics