Simple Unmark Launches Tool to Strip AI Watermarks and Hidden Characters From Text
haltakov · x · 2026-11-19
A new tool called Simple Unmark claims to remove AI watermarks from text by first stripping hidden Unicode characters, then rewriting word choices so statistical watermarks (SynthID, Claude watermarking) lose their signal while preserving meaning.
- Workflow: paste text → browser scans for hidden characters as you type → hidden chars removed and wording rewritten → review and copy.
- Privacy modes: Private mode (default) doesn't store text; upcoming Confidential mode has the browser verify and encrypt to the cleaning service, though DeepInfra still processes text at 2x credits.
- Pricing: 3 free cleans, then credits from $5, no subscription, up to 5,000 words per clean.
- Caveat: the author admits rewriting requires review and no tool can guarantee defeating every watermark detector.
Related event: Indie Dev Launches Simple Unmark to Strip AI Text Watermarks Locally(3 posts)→
More from Safety
- Lawyer banned two years after AI-drafted filings cited nonexistent court cases — GaryMarcus · 2026-11-19
- US blocks Chinese firms from renting Nvidia chips abroad while Microsoft nears $500M OpenAI deal for ByteDance — kevinnbass · 2026-11-19
- Recovering gambling addict slams Kalshi for bypassing state protections at 15x suicide rate — paulnovosad · 2026-11-19
- Human Principal launches private beta: cryptographic proof a human is behind your AI agent — yenkel · 2026-11-19
- Black hat SEOs exploit Google-indexed public Docs to poison AI search results for VPN queries — gaganghotra_ · 2026-11-19
- Wispr Flow's iPhone notetaker now records in-person client talks, consent debates follow — rcleffke · 2026-11-19