About the Em Dash Remover
An em dash remover deletes or replaces the em dash, the punctuation mark Unicode encodes as U+2014, in text you paste into it. The mark is the width of a capital M, which is where its name comes from, and it breaks a sentence harder than a comma and more softly than a full stop. It arrives in a document by three routes: a language model writes it, a word processor creates it by autocorrecting two hyphens, or it travels in on a paste from somewhere that already had it.
What is an em dash, and why does it end up in your text?
An em dash is the punctuation mark encoded at U+2014 in the Unicode standard, drawn roughly the width of a capital M. It separates a clause more sharply than a comma, and unlike a full stop it keeps the two halves of the thought inside one sentence.
Three things put one in a document. A language model writes it, because the mark appears heavily in current model output. A word processor makes it: both Microsoft Word and Google Docs replace two hyphens with a single em dash as you type, under an autocorrect rule that is on by default. Or the text carries it in from somewhere else, which is why em dashes turn up in copy pasted out of a CMS, a PDF or a chat window.
Knowing which of the three produced yours decides what to do about it, and the difference between the marks matters before you replace one.
Em dash, en dash or hyphen — which character is which?
Three different characters look similar and do different jobs. The list below names each one, gives its Unicode codepoint, and shows it in use.
- Hyphen — U+002D
- Joins words and splits them across lines. It is the key on the keyboard. Example:
a well-written brief. - En dash — U+2013
- Spans a range, roughly the width of a capital N. Example:
pages 41–58,Monday–Friday. - Em dash — U+2014
- Breaks a sentence, or brackets an aside in place of parentheses. Example:
the result — and nobody expected this — held.
The tool above treats the em dash and the en dash as separate characters for that reason: replacing an en dash inside a page range turns 41–58 into something that no longer reads as a range.
What should you replace an em dash with?
There is no single correct replacement, which is why the tool asks. The right mark depends on what the dash was doing in the sentence, and the four cases below are ordered by how often each one applies.
- A comma works where the dash brackets an aside. It is the safest default and the one the tool starts on.
- A full stop works where the dash joins two complete sentences that can stand alone.
- A hyphen works only where the destination cannot render anything better, such as a plain-text field or a legacy system.
- Deleting it works where the surrounding words already carry the break, and it is the option that needs the most care.
Deletion has one trap worth stating plainly: an em dash written without spaces around it sits directly between two words, so removing the character alone welds those words together. Leave the spacing option switched on and the tool inserts a single space in its place.
Why does AI writing contain so many em dashes?
Em dashes appear at a high rate in the output of current language models, which is the reason so many readers now treat the mark as a sign that a machine wrote the sentence. That frequency is observable in the text itself.
What is not established, and is therefore not claimed here, is any intent behind it. There is no evidence that the mark is inserted as a watermark, a signature or a deliberate marker of any kind. A high frequency in output and a hidden marking system are different claims, and only the first one has anything behind it.
Can you stop em dashes before they appear?
Partly, and the available evidence disagrees about how well. The method costs nothing to try: put an instruction in the model's own personalisation or custom-instructions settings telling it to write without em dashes, rather than asking in each message.
Three sources ranking on this subject reach three different conclusions, and the disagreement is worth having rather than hiding. Google's own AI Overview for the query hedges, saying opinions are mixed on whether prompt instructions stop the behaviour completely. The most-discussed Reddit thread on the term opens by stating that prompting to remove em dashes does not work, and recommends running a scripted replacement over the output instead. A ranking video claims a settings change fixed it permanently.
The practical reading: set the instruction, because it is free and it reduces the number, then check the output anyway. That is what this page is for.
How do you remove em dashes without this page?
Find and Replace does the job inside a document. Press Ctrl+H in Word on Windows or Cmd+Shift+H in Word on a Mac, put the em dash character in the search field, and put your replacement in the second field. Google Docs offers the same under Edit, then Find and replace.
Stopping the character being created is better than removing it afterwards. In Word the rule lives under AutoCorrect Options, in the AutoFormat As You Type tab, as the setting that converts hyphens to dashes. Google Docs keeps the equivalent under Tools, then Preferences, as automatic substitution.
Those methods cover one document at a time, they cannot make the spacing decision for you, and they do nothing about the invisible characters that usually arrive in the same paste.
When should you keep an em dash?
Keep the ones you put there on purpose. The em dash is a legitimate mark with a long editorial history, and writers use it deliberately for a break that a comma is too weak to carry. Replacing every instance flattens a rhythm the writer chose.
The rule worth applying: strip them from text you did not write and are about to publish under your own name, and leave them where you placed them yourself. A tool that treats the character as a defect in all cases is answering a narrower question than most people are asking.
What does this tool not do?
It does not make text pass an AI detector. Detectors score statistical properties of writing — sentence-length variance, vocabulary distribution, the rhythm across a passage — and swapping one punctuation mark for another leaves every one of those properties where it was.
It does not rewrite, edit or check grammar. The words that go in are the words that come out, with the dash characters changed and nothing else touched.
It does not know which of your dashes were deliberate. That judgement stays with you, which is the reason the section above exists.
It does not remove invisible characters or markdown. Those arrive in the same paste and are different jobs: the invisible character detector shows what is hidden in the text, and the markdown stripper handles the asterisks and hashes.