Zero-width and invisible
SOFT HYPHEN (U+00AD)
A conditional hyphen. It is invisible until the line breaks at that exact spot, at which point a hyphen appears.
- Codepoint
U+00AD- Category
- Zero-width and invisible
- Removed by default
- Yes
- Governing option
- No option needed
- Can be load-bearing
- No
Where it comes from
Word processors and typesetting systems that hyphenate automatically write them into the text. German and Dutch documents collect them by the hundred. They also come from PDF-to-text extraction, where the hyphen at a line end is preserved as U+00AD.
Why it matters
It sits inside words, so it silently breaks search, sorting, spell checking and deduplication: "Silbentrennung" is not "Silbentrennung". Because one is expected inside long words, it hides better in German or Finnish text than a zero width space ever could.
Before and after
The soft hyphen only draws itself if the line happens to break there. Otherwise it is a hole in the word that only software can see.
The character is shown as a labelled chip so you can see where it sits. In your text it draws nothing at all.
When it is legitimate
Inside a long compound word in a justified column, where you want to control exactly where hyphenation may happen. On the web, browsers hyphenate with CSS instead, and the character is usually left over from an older workflow.
How to remove it
The default pass removes it. If you actually need the hyphenation hints, keep the source document and clean only the exported text.
Is it an AI watermark?
Not a watermark
No. Soft hyphens come from typesetting and from PDF extraction. A model that was trained on hyphenated text can reproduce one, which makes it evidence of a document pipeline, not of authorship.
No connection to machine-generated text. It comes from software, from a keyboard, or from a person.
Questions
Why can I not see it in my editor?
Because it renders as nothing unless the line breaks at that position. Most editors show it only with "show invisibles" enabled.
Why is my text full of them after copying out of a PDF?
The PDF was hyphenated at layout time and the extractor preserved each break as a soft hyphen. Cleaning them restores whole words.
Check your own text
Paste anything into the cleaner to see every hidden character it contains, with position, codepoint and what happens to each one. Nothing is uploaded.