Blog
Honest guides to AI watermarks & provenance
Plain-language explainers on SynthID, C2PA, provider marks, and text signals — how removal actually works, where the limits are, and why we never claim “undetectable.”
All articles
How to remove SynthID watermarks (and why nobody can promise 100%)
A practical, honest guide to disrupting SynthID-class embedded watermarks on AI images — what actually works, what the limits are, and why residual risk always stays on the table.
- Provenance
How to remove C2PA / Content Credentials from an image
C2PA Content Credentials are a structured provenance record — not a pixel watermark. Here's what they are, how a structural strip works, and where the limits are.
- Provider marks
Removing the Gemini & Nano Banana sparkle watermark
Gemini-class exports carry a visible sparkle mark plus container provenance and possible embedded signals. Here's how to think about each layer — and why seamless removal is never guaranteed.
- Comparison
Why AI Stripped beats one-click, browser-only removers
Most watermark removers are single-layer and market “clean” results they can't verify. Here's the honest case for a multi-layer, account-gated pipeline that reports residual risk instead of hiding it.
- Text
Do AI text watermarks exist — and can you remove them?
SynthID-Text and statistical detectors work very differently from image watermarks. Here's how AI text signals work, what humanization can and can't do, and where the honest limits are.
- Fundamentals
AI provenance signals explained: metadata, C2PA & watermarks
Metadata, C2PA, visible marks, and embedded watermarks are four different signals with four different removal stories. This is the map that makes every other guide make sense.
Process a file the honest way
Multi-layer, best-effort, with an action report that tells you what changed and what didn't.