
Research
/Security News
OpenAPI React Query Codegen Compromised in Mini Shai-Hulud npm Supply Chain Attack
Ten malicious OpenAPI React Query Codegen versions were published to npm in the Mini Shai-Hulud attack, all with valid provenance.
ZeroMark.ai - clean AI text
zeromark-ai
Remove invisible Unicode watermarks and assistant chatter from AI text. Runs entirely in your browser: nothing is uploaded.
Clean AI text where you write it. Right-click a selection on any page, or press Alt+Shift+Z, and zeromark strips invisible Unicode watermarks, model scaffolding, and assistant chatter in place. Nothing is uploaded: this add-on has no network code.
Language models leave marks on their output. Some are cosmetic, like curly quotes and em dashes. Some are structural, like a leftover reasoning block or a citation marker left in a copy-paste. And some are invisible: zero-width spaces, bidi controls, and characters from the Unicode Tags block, which can carry an entire tracking ID or URL through a paragraph without rendering a single pixel.
What it removes
It shows its work
Every clean reports how many edits it made, and the sidebar names each hidden character it found by codepoint, so you can audit the change rather than trust it. All six rule groups are yours to switch off.
Privacy
Your text never leaves your browser. There is no server, no account, no analytics, and no network code of any kind in this add-on. It requests no access to any website: the cleaner is injected into the current tab only after you invoke it, which is why the install prompt asks for no permission to read or change your data on any site, and states that no data is collected.
The only thing it stores is your six rule toggles and your writing language.
Emoji that mean something are protected: a family or skin-toned emoji keeps the zero-width joiners that hold its parts together, so it stays one glyph instead of collapsing into three, while a stray joiner used as a watermark is removed.
Requirements
Firefox 140 or newer, on desktop. Firefox for Android is not supported: the two main entry points are a context menu and a keyboard shortcut, and neither of them exists there.
Same engine as the free cleaner at zeromark.ai, so both give identical results.
FAQs
Remove invisible Unicode watermarks and assistant chatter from AI text. Runs entirely in your browser: nothing is uploaded.

Research
/Security News
Ten malicious OpenAPI React Query Codegen versions were published to npm in the Mini Shai-Hulud attack, all with valid provenance.

Security News
Socket joins more than 100 technology, cybersecurity, and financial organizations calling for a global surge in cyber defense.

Product
Enterprise security teams can now detect malware, credential theft, suspicious network activity, and risky updates across Microsoft Edge extensions.