What this tool finds
Text copied from chatbots, PDFs, web pages and word processors often carries characters you cannot see: zero-width spaces and joiners, byte-order marks, soft hyphens, non-breaking and other unusual spaces, and bidirectional control marks. They can break search and replace, add stray gaps in a CMS, throw off character counts, cause odd line wraps, and trip validators and code. The scanner counts each kind, shows exactly where it sits, and returns a cleaned copy.
About hidden characters and AI detection
There is no reliable, publicly documented invisible watermark you can strip from chatbot text, and most AI text detectors judge writing patterns rather than hidden characters. So cleaning invisible characters is a hygiene step, not a way to hide how a text was made, and it will not change what a detector concludes. Em dashes, curly quotes and long ellipses are ordinary punctuation that many people and programs produce, so they are only reported unless you choose to replace them.
Safe defaults
Removing zero-width and direction characters and normalizing unusual spaces is safe for most English text. Two cautions apply. Zero-width joiners are part of how some emoji and several scripts such as Arabic and Indic languages are written, so the tool can keep joiners that sit between emoji, but for text in those scripts review the result before replacing your original. Direction marks matter in mixed left-to-right and right-to-left text, so leave that option off if you write in both.
Common uses
- Cleaning pasted text before it goes into a CMS, spreadsheet or code editor.
- Finding why two seemingly identical strings do not match.
- Removing non-breaking spaces that stop lines wrapping.
- Standardizing quotes and dashes to a house style.