Anthropic, OpenAI, Google, Meta, Microsoft, and Mistral have signed the EU Code of Practice on Transparency of AI-Generated Content, joining roughly 190 total signatory organizations. xAI did not sign. The code is tied to EU AI Act Article 50 transparency requirements; the Transparency Code took effect August 2, 2026, requiring AI companies to mark AI-generated or edited content in a way other systems can identify.
Anthropicís implementation encodes watermarks as statistical signals directly into token choices, designed to survive copy-paste and light editing. The marking applies globally across all Claude products, the API, Claude Code, and major cloud platforms, not only in the EU. Claude models launched in the EU on or after August 2, 2026 support marking at launch. For image outputs, SVG, PNG, and JPG files use signed C2PA provenance metadata to indicate Claude processing and detect metadata tampering.
Anthropichas not yet released a watermark detection tool, which the EU code calls for. The company states that a detected watermark cannot prove Claude originally authored content, since Claude may have edited, translated, or summarized human-created material, and that missing marks are inconclusive given factors such as heavy rewriting, short passages, older models, unsupported surfaces, or stripped file metadata.
OpenAI updated its support page on text watermarking nine days before the signing was publicly reported.