No account neededRuns in your browser: your text is never sent to Awiser
AI text cleaner
Paste text from ChatGPT, Claude or Gemini. The cleaner strips hidden characters, citation codes and markdown symbols, and swaps typographic dashes and quotes for plain ones. Your wording stays as it is.
Up to 200,000 characters.
Clean text
The clean text appears here
Paste your text, or see what the cleaner does to a copied chat answer.
What to clean
What the cleaner fixes
- Hidden characters
- Zero-width spaces (U+200B), joiners (U+200C, U+200D), word joiners (U+2060), byte order marks (U+FEFF), soft hyphens, direction marks and Unicode tag characters. They take no room on screen, yet they split words for search, break code and can carry hidden text.
- Special spaces
- No-break (U+00A0), narrow no-break (U+202F), thin and other typographic spaces become plain spaces, and line separators become line breaks. They look like spaces, but they stop a line from wrapping and make find-and-replace miss a phrase.
- Em and en dashes
- Em dashes (U+2014) and en dashes (U+2013) become hyphens. Between two words a space stays on each side, so a dash never glues them into one; between numbers it reads as a range. Non-breaking hyphens (U+2011) become plain ones.
- Curly quotes
- Curly quotes and apostrophes become straight ones, which code, spreadsheets and search expect.
- Ellipsis character
- The single ellipsis character (U+2026) becomes three full stops.
- Markdown symbols
- Heading marks, the asterisks around bold and italics, asterisk list marks and horizontal rules go, and a link becomes its text with the address in brackets. Code blocks are left as they are.
- Citation codes
- Codes such as citeturn0search0, left where a chat app showed a source link. Removing one does not bring the source back: if you need it, copy the link from the chat.
- Styled letters
- Bold, italic, script and wide Unicode letters, often pasted from fancy text generators, become plain letters. Screen readers spell them out one by one and most search boxes miss them.
- Lookalike letters
- Cyrillic or Greek letters drawn exactly like Latin ones, inside a word that is otherwise English. A word written in Russian or Greek is never touched.
- Extra spaces
- Spaces at the end of a line, doubled spaces between words, more than one empty line in a row, and blank lines at the start or the end. Indentation stays.
Where these characters come from
- Copy buttons copy the markdown. A chat app shows bold text and headings, but its copy button hands over the source, so the asterisks and hashes land in your email or post.
- Answers that searched the web carry citation codes. On screen they turn into source links; pasted anywhere else they stay as codes.
- Models write typographic characters. Some ChatGPT models have been reported writing narrow no-break spaces where a plain space belongs. They look identical, which is why nobody notices them.
- Hidden characters can carry instructions. Unicode tag characters are invisible to people but readable by AI models, which is how instructions get hidden in a text an assistant is asked to read.
Questions
Is my text sent to Awiser?
No. The text is cleaned in your browser and never sent to us. The page remembers which switches you turned off, in this browser only, and never stores the text.
Will cleaning make AI text undetectable?
No. The cleaner changes characters, not wording, and detectors that score a text read its wording, which stays exactly as it was. What cleaning does is stop hidden characters and leftovers from breaking your page, your code or your search.
Does it change my words?
No. It removes characters you cannot see and swaps typographic marks for plain ones. Letters change in two cases only, both marked in the result and both switchable: styled Unicode letters become plain, and a Cyrillic or Greek letter inside an English word becomes its Latin twin.
Are emoji and other languages safe?
Yes. Emoji keep the joiners that hold them together, accented letters are never touched, and the joiners and spaces that Arabic, Persian, Hindi, Thai or Chinese text needs stay in place. Direction marks are kept whenever the text holds Hebrew or Arabic.
What do the marks in the clean text mean?
A shaded character is one the cleaner changed, and a thin bar shows where it removed something you could not see. Point at either to see what was there. The marks are never copied with the text.
Related tools
- AI crawler checker
Check whether ChatGPT, Claude and Perplexity may read a page under your robots.txt.
Awiser is the builder network where founders show their skills, find collaborators, and grow their projects with an AI copilot. More mini-apps