Identify and remove hidden Unicode artifacts, clean up Markdown paste residue, and restructure Gemini draft text with transparent, browser-based inspection.
See it in action
Review how invisible Unicode characters and token distribution watermarks like SynthID-Text affect generated passages.
Sure! Here’s a summary — **Key points** - First item > Note: important
Sure! Here's a summary — Key points • First item Note: important
Why
Identify and remove hidden Unicode artifacts, clean up Markdown paste residue, and restructure Gemini draft text with transparent, browser-based inspection.
Direct copy-pasting from Gemini web or workspace interfaces often introduces zero-width spaces, narrow non-breaking spaces (U+202F), unrendered Markdown tags, and trailing metadata blocks. These are deterministic byte-level artifacts that can be fully detected and stripped locally in your browser.
Statistical watermarking concepts, like Google's SynthID-Text research, operate at the token distribution layer during inference using secure pseudo-random keys. These properties cannot be observed or validated through client-side string analysis, nor can any tool guarantee statistical watermarking removal without comprehensive semantic rewriting.
Invisible-character and Markdown cleaning run locally in your browser. We don’t store Free-tool text.
How
Paste your raw text into the input field. Our client-side analyzer instantly inspects the string against 60 invisible Unicode characters, highlighting exact character offsets and code point values.
Click Clean to strip unwanted zero-width characters, normalize irregular spaces (like U+202F), and strip broken Markdown tags entirely within your local browser sandbox.
If you need natural variation in phrasing and syntax, use our paid reconstruction and humanizer engines to rewrite the passage while maintaining original factual accuracy.
Features
Detects zero-width spaces, NNBSP, BOM, directional marks and 60+ other invisible code points without rewriting your words.
Strip headings, lists, quotes, and emphasis markers into plain text ready for CMS and editorial review.
Pro rewrite regenerates meaning-preserving text when a watermark is baked into wording — not just hidden Unicode.
Rewrite AI flavor into more natural cadence. No tool can guarantee detector bypass; results vary.
Definition
When copying text from Google Gemini interfaces, copied snippets frequently contain invisible Unicode characters, non-standard whitespace such as narrow no-break spaces (U+202F), and Markdown syntax fragments. These character-level artifacts can break code interpreters, fail strict formatting parsers, or trigger basic heuristic filters.
Beyond character-level residue, Google researchers published landmark findings on SynthID-Text in Nature, detailing pseudorandom sampling adjustments during token generation. It is critical to distinguish these statistical watermarking frameworks from invisible string markers: statistical generation marks modify word probability distributions rather than appending hidden characters.
While Anthropic's public disclosures for Claude confirmed that generation marks do not insert hidden characters into output strings, Gemini's underlying generative research belongs to a distinct technical implementation family. This independent utility is not affiliated with Google and does not claim to inspect, decrypt, or break Google's proprietary keyed statistical watermarks.
Our free tool operates strictly on observable string data in your browser. It highlights exact invisible code points and strips Markdown residue. For users seeking stylistic variation, our paid reconstruction models offer semantic restructuring, though we make no claims of bypassing third-party detectors like Turnitin, GPTZero, or Pangram.
Sources
Hidden Unicode, Markdown residue, and Claude’s official statistical watermark are three mechanisms. This page cites the public specs; it does not claim detector bypass.
| Mechanism | What it is | What this site does |
|---|---|---|
| Invisible Unicode | Zero-width and format code points (including U+200B and U+202F) that can ride along in a paste. | Free local scan and delete. Does not rewrite sentences. |
| Markdown / rich-text residue | Hashes, fences, emphasis markers, and copied rich-text spacing. | Free local strip to plain text. |
| Statistical wording watermark | A sampling pattern in word choice. Official Claude text watermarks add no hidden characters. | Pro meaning-preserving rewrite. Not a detector score, not a guarantee. |
“Nothing is added to the text and there are no hidden characters.”
“U+200B ZERO WIDTH SPACE is intended for line break control without displaying a visible space.”
Markdown tokens follow the CommonMark specification: CommonMark spec.
Use cases
Sanitize drafts before WordPress, Webflow, or a headless CMS — without dragging invisible characters into meta fields.
Open the tool →
Turn chatty AI Markdown into clean plain text for writers, fact-checkers, and style guides.
Open the tool →
Ship client-ready copy faster: remove paste residue locally, then upgrade only when you need rewrite-based cleaning.
Open the tool →
Normalize assistant output before docs, changelogs, and internal knowledge bases — with transparent detections.
Open the tool →Testimonials
“The free scan lists the code points. I clean the paste in the browser, then only use Pro if the wording itself needs a rewrite.”
Content ops
SEO & CMS
“Markdown from chat was breaking our editor. Stripping hashes and fences locally is the whole job — no file upload.”
Editorial
Publishing
“Hidden Unicode broke string equality in a parser. Character cleanup first; sentence rewrite is a different layer.”
Engineering
Docs & support
Tools
Dedicated landing pages for the searches people actually type.
Understand how Anthropic marks AI text at generation time, strip copy-paste formatting residue in your browser, and reconstruct wording with clean syntax.
Claude guide →Inspect and strip U+202F narrow no-break spaces, invisible Unicode residue, and Markdown artifacts from ChatGPT pastes locally, or use Pro rewrite for wording-level adjustments.
ChatGPT guide →Inspect and list 60 invisible Unicode characters, U+202F narrow spaces, and markdown formatting artifacts in pasted text with transparent, honest reporting.
AI Watermark DetectorPricing
Browser-based paste cleaning is always free, while our pricing page offers Pro credits for deep text rewrites.
See pricingThis tool is a character sanitizer and text editor, not an evasion service. We inspect and eliminate 60 invisible Unicode code points and structural markup. We do not provide cryptographic validation against Google's private SynthID keys, nor do we promise evasion of commercial AI classifiers such as GPTZero, Turnitin, or Pangram. We recommend manual review for all academic, legal, and professional publications.
FAQ
Unicode residue consists of literal invisible characters (such as U+202F, zero-width spaces, or soft hyphens) left behind by web copy-paste operations. Statistical watermarks, like those described in Google's SynthID-Text research, are mathematical patterns embedded in token selection during model generation. Invisible characters can be detected and stripped completely in your browser, whereas statistical watermarks contain no hidden characters and require a secret cryptographic key to verify.
Our 60-character invisible Unicode scanner, Markdown sanitizer, and local text cleaning tools are completely free and run directly in your browser. Advanced meaning-preserving reconstruction, tone humanization, and full structural paraphrasing require paid credits to cover backend inference computing costs.
Free Unicode scanning and formatting cleanup execute 100% locally on your device via JavaScript; no text is sent to our servers. When using paid paraphrasing features, your text is processed ephemerally over encrypted TLS connections to generate the rewritten text and is never used to train machine learning models or stored permanently.
No. We do not guarantee passing scores on AI detection platforms such as Turnitin, GPTZero, CopyLeaks, or Pangram. Commercial classifiers use volatile statistical heuristics that frequently produce false positives and false negatives. Our goal is to provide clean, error-free text and natural paraphrasing rather than evasion guarantees.
You should avoid automated paraphrasing on technical source code, mathematical equations, legal contracts, medical guidance, or strict factual citations where unintended word substitutions could compromise technical correctness or legal validity. Use the free Unicode cleaner instead to sanitize characters without changing phrasing.
No. Google's SynthID-Text system relies on private cryptographic keys and scoring algorithms held by Google to evaluate statistical token distributions. Because this site is an independent utility with no access to Google's proprietary keys or internal infrastructure, no third-party tool can definitively verify or measure SynthID presence.
About us
Our free tool operates strictly on observable string data in your browser. It highlights exact invisible code points and strips Markdown residue. For users seeking stylistic variation, our paid reconstruction models offer semantic restructuring, though we make no claims of bypassing third-party detectors like Turnitin, GPTZero, or Pangram.
Identify and remove hidden Unicode artifacts, clean up Markdown paste residue, and restructure Gemini draft text with transparent, browser-based inspection.