The job
What the em dash replacer does to a draft
You paste a draft, you name a replacement, and every long dash in the text becomes that replacement. The em dash replacer matches two codepoints: U+2014, the em dash, and U+2013, the shorter en dash used in ranges. One pass covers both.
The replacement is typed, not chosen. Competing pages hand you a menu of four to six options and stop there. This is a text field, so the string can be a hyphen, a comma with a space after it, a double hyphen, a spaced hyphen, or a word. Whatever you can type, it will write.
Nothing is uploaded and nothing is metered. The run happens in the page, your pasted text stays in the left box, and the rewritten version appears on the right for copying. In a drafting workflow that matters: try a comma, read the result, try a colon instead, as often as you like.
The run
How to replace em dashes in three steps
Under a minute from paste to copy, and repeatable while you decide what reads best.
- Step 1
- Paste the draft into the left box. The bar above shows a live character and word count, which is handy when you are writing to a length.
- Step 2
- Set the field marked Replace dashes with. It opens on a hyphen, and an empty field falls back to a hyphen rather than deleting anything.
- Step 3
- Press Replace dashes, read the right-hand box, then copy. To try a different replacement, change the field and press it again. The source text is still sitting there untouched.
Decide once
Pick the replacement before you paste
Because one string is applied everywhere, the decision is worth making before the run rather than after. Read two or three of the dashes in your draft and see which job they are doing. Most drafts use the mark for one purpose throughout, which is exactly why a single substitution usually works.
- Asides, in pairs
- A comma is the closest match. Type it with no trailing space if the draft already spaces its dashes.
- A wind-up before a conclusion
- A colon carries the same beat in a more formal register.
- Two complete statements joined
- A full stop plus a space is cleanest, and you re-capitalise the next word yourself afterwards.
- Anything a machine will parse
- A plain hyphen. This is the default in the field for a reason.
- A style guide that wants a spaced dash
- Type a space, a hyphen and a space. The field accepts the whole string.
Two characters
Both dash widths, one pass, one string
The en dash comes along for the ride. That is more than some tools on this term manage: one well known replacer page never mentions U+2013 at all, and another treats it as an extra checkbox next to smart quotes.
The limit is the flip side. Both characters receive the same replacement, so a draft using en dashes for date and page ranges and em dashes in prose cannot be sorted out in one go, and a comma inside a page range is worse than the dash was. Run the find and replace tool for one of them first, then bring the rest here.
Nothing else is matched. Typed hyphens stay, double hyphens stay, minus signs stay.
Sequencing
Where an em dash replacer fits in the workflow
Run it late, after the writing is settled but before the text enters anything that stores or renders it. Replacing dashes early means doing it again after the next round of model-assisted edits, because the model will put them back.
Run it before the text touches a system that will read it as data. A hyphen is one ASCII byte, while U+2014 is three bytes in UTF-8, and those three bytes only survive if every stage of the pipeline agrees on the encoding. When one stage does not, a dash is usually the single broken thing in an otherwise clean import.
- Before a CSV or a spreadsheet import
- Encoding mismatches show up here first, and a stray long dash in a cell is the classic symptom.
- Before a URL slug or a filename
- It has to be escaped, so the clean slug you wanted turns into a percent-encoded string.
- Before a config value or an API payload
- A plain field is the wrong place to hide a three-byte character.
- After the final model-assisted pass
- Anything earlier and the next rewrite reintroduces the mark.
Checks
What to check after the replacement runs
Two things reliably catch people out, and neither is a fault in the text.
First, spacing is not adjusted. The replacement goes exactly where the dash was, so a spaced dash swapped for a comma leaves a space before the comma. Plan the string to account for it, and if a run leaves doubled spaces around, the whitespace minifier tidies them in a second pass.
Second, the footer counter reports characters removed. A one-for-one swap removes nothing, so it reads Processing complete rather than showing a number. That is a finished run. Read the result box to confirm the change, not the counter.
Straight answer
Does swapping dashes fool an AI detector?
No, and nothing here promises a detector outcome. Detectors report signals and probabilities rather than proof of authorship, and a punctuation swap does not change who wrote a sentence.
The useful claim is smaller and it is about readers. Models produce the long dash at a rate most people do not, which is why three of them in a paragraph reads as machine-written at a glance. Replacing them removes that one visual pattern. The rhythm of the sentences and the shape of the paragraphs are exactly as they were, and those are the harder tells to edit out.
In practice
Who runs an em dash replacer, and when
Content teams clearing a batch of drafts into a house style that bans the mark. Students and writers who noticed the pattern in their own assisted draft and want it gone before submission. Newsletter and social writers whose platform renders the character unpredictably.
On the technical side: anyone moving prose into a field a program will read. Developers preparing seed data, analysts cleaning an export before it hits a spreadsheet, and CMS editors whose publishing chain has one stage that never quite agreed about UTF-8.