Sebastian Raschka By Sebastian Raschka, PhD

How Claude Watermarks AI-Generated Text

Read the original on Sebastian Raschka →

The article titled "How Claude Watermarks AI-Generated Text" presents a 48‑minute video walkthrough that explains the process of token sampling, watermark detection, and watermark removal in AI‑generated text. It provides a detailed, step‑by‑step demonstration of how these techniques are applied and how they can be reversed.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at Sebastian Raschka.

Towards Data Science
Sep 6

Text Watermarking in Python: Catch Whoever Copies Your Writing

The article explains how AI companies embed watermarks into billions of words daily and shows how readers can apply three main watermarking techniques to their own writing. It details practical steps for implementing these methods in Python and discusses experimental results that test which watermarks endure copy‑paste, editing, and paraphrasing. The post provides actionable guidance for writers seeking to protect their content from unauthorized duplication.

By Chien Vu Minh
arXiv AI
5d ago

Watermarking Diffusion Language Models

The paper introduces the first watermark designed specifically for diffusion language models (DLMs), which generate tokens in arbitrary order unlike traditional autoregressive models. It overcomes the challenge of missing prior tokens by applying the watermark in expectation over the context and promoting tokens that strengthen the watermark when used as context. Experiments show a >99% true positive rate with minimal quality loss and comparable robustness to existing autoregressive watermarks.

By Thibaud Gloaguen, Robin Staab, Nikola Jovanovi\'c, Martin Vechev
arXiv Computer Vision
Sep 7

AngelFingerprint: A Traceable, Explainable, and White-Box Stealthy Watermark for Text-Guided Image Editing

AngelFingerprint introduces a watermarking framework for text-guided image editing that embeds the editing prompt’s CLIP text embedding directly into the diffusion model’s weights via a LoRA module. The watermark is recoverable from image pixels alone, providing an explainable trace of the edit while remaining stealthy even under full white-box access. Experiments on the MagicBrush dataset show the extractor achieves 86% top‑1 accuracy in 200‑way prompt retrieval, outperforming prompt inversion methods.

By Bo-Han Kung, Futa Waseda, Ching-Chun Chang, Isao Echizen, Shang-Tse Chen