Anthropic's text watermarking reaches edited Claude output

Anthropic's new Claude text watermark can remain detectable after copied text is pasted elsewhere and can appear in translations, summaries and more extensively edited material, Business Insider reported. Anthropic says the global rollout is intended to comply with the EU AI Act because it cannot apply the feature by region. The report says lighter editing is less likely to trigger the signal, while users and developers have raised concerns about authorship and removal tools.