Anthropic Uses Watermarks to Meet EU Rules

American company Anthropic announced the implementation of invisible watermarks on the texts generated by its model "Claude," in compliance with the new European Union laws requiring the disclosure of AI-generated content. The existing marks within the text remain detectable when copying and pasting, but they may disappear in short texts or those that have undergone extensive editing. Additionally, the company will attach digitally signed metadata with media such as images to ensure transparency. This step is part of efforts to comply with the European AI law requirements, aiming to avoid fines, and the company plans to apply these measures globally starting August 2026, with previous models gradually becoming compatible. However, the company has not yet released an independent detection tool to verify the watermarks.

articleDetails.fontSize.sizeحجم النص

تنويه: هذا ملخص تم إنشاؤه بواسطة الذكاء الاصطناعي

عرض المقال بالكامل