synthid

Strip multi-vendor AI provenance marks: Unicode text hygiene, statistical rewrite hooks, and C2PA/metadata from PNG/JPEG/SVG/PDF/DOCX/HTML/MD

Remove visible and invisible AI watermarks and provenance metadata from images and video. Python library and CLI for SynthID, C2PA, EXIF, IPTC, XMP, and common generative-AI marks.

Remove invisible AI watermarks (SynthID, StableSignature, TreeRing) and strip AI metadata from images. Open-source CLI & Python toolkit.

Cross platform CLI tool for removing watermarks from images and videos generated by Google Gemini, Veo, and NotebookLM.

🍌 Remove invisible AI watermarks from Google Gemini images — npm package

AI watermark removal built for Apple Silicon. SynthID-class pixel regeneration, visible vendor labels and provenance metadata, with every pipeline decision measured on Metal.

Invisible, key-verifiable watermarks for AI/LLM-generated text — SynthID-Text-style, in Rust + WebAssembly (Node + browser). Live playground included.

A specialized tool for removing invisible AI watermarks, such as Google SynthID, from images.

Clean. Detect. Generate. In Seconds.

Strip EXIF, GPS, C2PA / Content Credentials, AI watermarks, and Stable Diffusion / ComfyUI / Midjourney / ChatGPT prompt blocks from any image. Three modes: safe (pixel-identical), paranoid (re-encode), nuclear (disrupt SynthID/Digimarc). CLI + macOS desktop app. 100% offline.

ComfyUI custom nodes for detecting provenance marks, removing visible and invisible AI watermarks, stripping metadata, and erasing selected regions.

Remove invisible AI watermarks and manage metadata from images generated by popular AI models like DALL-E, Midjourney, and Stable Diffusion.

AI watermarker for images and text that replicates SynthID's functionality.

Research demo: scrub generative text watermarks (e.g. SynthID-Text) via H-token sliding-window MLM token replacement. Gradio UI, multilingual (ModernBERT / mmBERT).

Find and remove AI provenance signals from files you own: hidden Unicode, token watermarks, visible image marks, and C2PA/EXIF/XMP metadata. Zero-dependency Python 3.10+ core; models, GPUs, and system binaries sit behind opt-in adapters.

An invisible watermark is still a signal. A lightweight CNN ensemble that detects SynthID — the imperceptible watermark in ChatGPT / GPT-Image-2 images — trained on a laptop and validated against the official verifier.

Research-only AI watermark & provenance robustness toolkit: local reverse proxy (OpenAI/Anthropic/Gemini) + CLI stripping C2PA/EXIF/XMP, zero-width & homoglyph Unicode, KGW text watermarks, DWT/Tree-Ring image stego, AudioSeal, PDF/DOCX/PPTX/XLSX metadata, and Trojan Source (CVE-2021-42574) code scanning.

数字水印实验室：AI 文本水印检测、C2PA 内容凭证读取、零宽字符隐写、LSB 与频域盲水印，9 个纯前端工具，打开即用不上传文件

Check any text for the traces AI chat interfaces leave behind — Claude HTML class names, zero-width characters, exotic Unicode spaces. Zero dependencies, MIT, runs offline.