pagedMark Tool Released: Removes Invisible AI Watermarks from Images and Video
d0ofz · reddit · 2026-08-19
The author released pagedMark, a tool designed to remove AI provenance from self-generated content.
Features:
- Dual Handling: Removes metadata (C2PA, EXIF) and pixel-embedded invisible watermarks (e.g., SynthID).
- Mechanism: Regenerates the image to strip watermarks, meaning the output isn't identical (faces/text may change), aiming to remove the signal while maintaining similarity.
- Support: Covers invisible marks from ChatGPT, GPT-Image, Z-Image Turbo, and visible marks/metadata from Sora, Veo, Kling, etc.
- Platform Optimization: Memory-aware processing for Apple Silicon to avoid swap slowdowns.
The author claims that after removing GPT-Image watermarks, OpenAI's verification tool shows 0% AI detection.
More from Fun
- Critique: Using AI to Generate "Devil's Advocate" Comments on LinkedIn — lilyraynyc · 2026-08-19
- NeurIPS 2026 Rebuttal Phase in a Nutshell: A Relatable Meme — CSProfKGD · 2026-08-19
- OpenRouter down? It turns out Stripe shut off their service — matt_slotnick · 2026-08-19
- Meme: OpenAI employee vs Anthropic employee posting styles on the timeline — tekbog · 2026-08-19
- Redditors share humanoid robots prepping for the WHRG26 robot games — BigBangFlash · 2026-08-19
- How 'models' were quantised in the 19th century: British Library dataset — vanstriendaniel · 2026-08-19