Wire · regulatory
Anthropic To Mark Claude Text & Files Under EU AI Act Code
◆ Sectors
◆ Geography
◆ Source
◆ Verified
Fusion42 · 11 August 2026 · Fusion42 review
Anthropic commits to embedding invisible watermarks in AI-generated text and adding signed provenance metadata to files produced by its Claude models worldwide, complying with the EU AI Act's transparency code. This aims to help detect AI processing in content, although marks can be lost through heavy editing or using older models.
This Wire brief sits within Fusion42's coverage of AI & ML, and 14 sources have reported it between 11 Aug 2026 and 23 Aug 2026.
◆ ◆ The Wire takeaway
EU AI transparency rules now require you to prove AI involvement in content via watermarking. If your product handles generative AI text, new compliance and detection requirements affect how you label and track outputs.
◆ Coverage
14 sources · first reported 11 Aug 2026 · latest 23 Aug 2026
◆ Related on Wire
◆ Topics