Anthropic introduced a watermarking system for its AI chatbot Claude in August 2026 to meet the European Union AI Act’s transparency rules for AI-generated content [1, 2, 3]. The watermarking method uses a version of Google DeepMind’s open-source SynthID-Text technology, which was developed in 2024 and previously adopted by Google’s Gemini chatbot [2, 3].
The watermark embeds a detectable pattern in the AI-generated text by subtly influencing low-stakes word choice probabilities during generation, without altering the text’s meaning or quality [1, 2, 3]. Anthropic emphasized that the watermark is invisible to readers and does not add extra characters, tokens, or cost to producing responses [1, 3]. A company spokesperson said, "Watermarking does not impact the quality of Claude’s output. To a reader, a watermarked response is indistinguishable from an unwatermarked one" [2].
Unlike AI detection methods that analyze writing style for AI ‘tells’, the watermark can only be identified with a specific key encoding the pattern and contains no user or chat-identifying information [1, 2]. Light edits to watermarked content may reduce but not fully erase the watermark, while a complete rewrite replacing every word would remove it but also change the text’s AI origin [2]. Anthropic stated, "Light editing probably won’t remove the watermark completely, while a complete rewrite where every word is replaced will. In the latter case, of course, it’s arguable whether the text can any longer be described as AI-generated" [2].
Anthropic plans to release an API for users to check if text is watermarked by Claude [2]. Other major AI providers serving the EU market are also implementing watermarking to comply with the AI Act, though OpenAI has yet to detail its plans for ChatGPT [1, 3].
The EU AI Act’s requirement to embed detectable watermarks in AI-generated text took effect on August 2, 2026 [1]. Anthropic published a detailed blog post on August 14 explaining its approach, and media outlets reported on the technology shift on August 15 [2, 3]. Some Claude users have criticized the watermark system on social media, with a few canceling subscriptions over concerns about transparency and AI content tracking [2]. A Reddit user quoted by TechCrunch stated, "The only reason you wouldn’t want this is to lie to people" [2].
Anthropic’s rollout of watermarks in Claude marks a concrete step following the EU’s new regulations. The company’s planned detection API is expected to launch soon, providing a way for users and regulators to verify the AI origin of text.
