AI text cleaner
Replace em dashes, straighten curly quotes and ellipses, and remove the invisible characters that travel with text copied out of a chatbot.
How to use ai text cleaner
Beyond Markdown, AI output carries a typographic accent: em dashes, curly quotes, true ellipsis characters, and the odd invisible mark. This tool normalises them to what a keyboard produces, with a choice of what each em dash becomes.
- 1Paste the AI-written text
The counter under the editor reports how many AI artifacts are present — special dashes, curly quotes, ellipses, and hidden characters — which tells you before cleaning whether this text needs the pass at all.
- 2Pick a dash policy and the other switches
Em dashes offers a spaced hyphen, a comma, a bare space, or keeping them as they are. Straighten quotes, Ellipsis to three dots, and Remove hidden characters are independent switches, so you can straighten quotes for a config file while leaving the dashes alone.
- 3Clean and use the neutral text
Select Clean text. Ranges like 12–15% keep a tight hyphen rather than picking up your dash filler, quotes come out straight, and the result pastes safely into code, forms, CSV files, and CMS fields.
When this tool is useful
The conversions, and what is deliberately spared
Dashes are position-aware: an em or en dash between digits is a range, so "12–15%" becomes "12-15%" with a tight hyphen no matter which filler you chose, while dashes between words take the filler — spaced hyphen, comma, or space. Curly single and double quotes, including the low-9 and prime variants, become straight quotes, and the single … character becomes three full stops.
The hidden-character pass removes zero-width marks, soft hyphens, word joiners, directional controls, and byte-order marks, and converts the exotic spaces — non-breaking, thin, figure — into ordinary ones. These are the characters that make AI-copied text fail searches, break form validation, and corrupt diffs, and the counter under the editor shows how many were found.
Normalising typography does not disguise AI text, and this tool does not claim to: detectors analyse word choice and sentence rhythm, which punctuation changes do not touch. The honest reasons to normalise are practical — a house style that wants straight quotes, a parser that chokes on curly ones, an import that mangles non-ASCII, a search that keeps missing.
AI text cleaner questions
Assistants are trained on published, professionally typeset prose, where the em dash is standard. Keyboards have no em dash key, so typed text mostly uses commas and hyphens instead — which is why the contrast is noticeable.
It depends on the destination. A comma reads most naturally in plain prose; the spaced hyphen keeps the dash’s pause and is the default; a bare space just removes the mark. Dashes between numbers are exempt from all three — a range always gets a tight hyphen.
Parsers expect the straight quote characters, and a curly quote is a different character entirely. A pasted snippet then fails with a syntax error that is nearly invisible to the eye. Straightening quotes is the single most valuable conversion for anything technical.
Yes. The variation selector that renders emoji in colour is deliberately kept, and so are the zero-width joiners that hold composed emoji together — families, professions, and skin-tone sequences stay whole. Only joiners with no emoji or non-Latin letters on both sides are treated as debris.
No, and that is not what it is for. Detectors look at wording patterns, not punctuation, so normalising typography changes how text looks rather than how it profiles. The tool fixes practical problems: style compliance, code and data safety, and search matching.