Invisible Character Remover.

Invisible characters span several Unicode categories and serve various technical purposes — some legitimate, some problematic when they appear unexpectedly in text.

Zero-width characters (U+200B, U+200C, U+200D, U+2060, U+FEFF, and others): Used in some scripts for typographic control. Problematic in general text because they are invisible, break string matching, and are used in AI text watermarking.

ASCII control characters (U+0000–U+001F except tab, newline, carriage return): The first 32 characters of the ASCII standard are control codes. Most have no visual representation in modern text. Null characters (U+0000), form feeds, vertical tabs, and delete characters (U+007F) appear in text occasionally as artifacts and can cause rendering problems.

Bidirectional control characters (U+202A–U+202E, U+2066–U+2069): Control the direction of text rendering for right-to-left scripts. When injected into text that doesn't need them, they are invisible but can reverse character display order in certain contexts — a known security vulnerability called the Trojan Source attack in code.

Soft hyphen (U+00AD): Invisible in most contexts, only becoming a hyphen when a line needs to break at that position. Often appears in text exported from typesetting software.

Non-breaking spaces (U+00A0) and variations: These are visible as spaces but are distinct from standard spaces and can cause word-wrap and string-matching issues. This tool optionally converts them to standard spaces.

Invisible characters create three categories of problems in professional text workflows.

Technical failures: String matching, find-and-replace, regular expressions, and programmatic text processing all operate at the character level. Two strings that look identical can fail string equality checks if one contains invisible characters the other does not. This causes failures in search, indexing, database lookups, and any system that compares text values.

AI detection flags: AI detection platforms scan for invisible Unicode characters as explicit watermarking signals. Their presence increases AI detection scores. For any content that will be reviewed by automated AI detection, invisible characters are a direct liability.

Rendering anomalies: Some invisible characters affect text rendering in specific environments. Bidirectional control characters can reverse text display in browsers and email clients. Zero-width joiners affect character combining in certain scripts. Null characters truncate strings in some systems. Soft hyphens create unexpected line breaks in some typesetting contexts.

Cleaning invisible characters from text before it enters any professional workflow is defensive practice regardless of AI concerns. The characters create problems across a wide range of systems and their presence in text is almost always accidental or unwanted.

Paste any text into the input field. The tool runs a character-by-character scan comparing each code point against the full list of invisible and non-printing character categories.

Characters removed by this tool: - All zero-width Unicode characters (U+200B, U+200C, U+200D, U+200E, U+200F, U+202A–U+202E, U+2060–U+2064, U+206A–U+206F, U+FEFF) - ASCII control characters (U+0000–U+0008, U+000B, U+000C, U+000E–U+001F, U+007F) - Soft hyphen (U+00AD) - Combining Grapheme Joiner (U+034F) - Other Unicode format characters that produce no visible output

Characters preserved: - Standard space (U+0020) - Tab (U+0009) - Newline (U+000A) and carriage return (U+000D) - All visible characters across all scripts - Standard punctuation and symbols

The removed count shows how many invisible characters were found. Clean text produces a count of zero.

AI-generated text from ChatGPT, Claude, Gemini, and Grok has been documented containing invisible characters by researchers who analyzed outputs at the Unicode code point level. The distribution of these characters — particularly zero-width spaces and directional marks — is frequently non-random, suggesting systematic insertion rather than incidental inclusion.

The use of invisible characters as text watermarks is a well-established technique in information security research predating AI language models. Digital watermarking of text through Unicode character insertion has been studied as a way to trace document leaks and identify origin. AI labs implementing similar techniques for content tracking would be following a documented playbook.

What is less clear is the specifics of each model's watermarking strategy. OpenAI, Google, Anthropic, and xAI have not published detailed disclosures about invisible character watermarking in their text outputs. What researchers can document is that the characters appear and that their distribution patterns differ from what would be expected in clean human-typed text.

This tool removes all of them regardless of source, model, or intent. For any AI-generated text entering a professional workflow, running it through this scanner is straightforward and costs nothing.

After processing with this tool, the removed count confirms how many invisible characters were found and removed. A count of zero on the first pass means the text was already clean. A positive count on re-processing the output would indicate the tool missed something — which should not occur, but if it does, re-running is instant.

For independent verification, you can paste the cleaned text into a Unicode inspector or character viewer to confirm no unexpected code points are present. Browser developer tools also allow inspection of text at the code point level in the console: Array.from("your text").map(c => c.codePointAt(0).toString(16)) will show the hex code point of every character.

If you need to verify that specific invisible characters are absent — for example, to confirm U+200B specifically is gone — a find-and-replace in a code editor that supports Unicode character search (VS Code, Sublime Text) can locate any specific code point.