According to TechCrunch, Anthropic published a blog post on Friday addressing external concerns about Claude's text watermarking mechanism. This move aims to comply with the transparency guidelines of the EU AI Act, which requires AI companies to adopt systems capable of identifying AI-generated content. Anthropic has adopted Google DeepMind's SynthID-Text solution, where the watermark is invisible to readers but detectable by those holding the key, without affecting the quality of Claude's output.