# Anthropic's Claude Will Now Add Invisible Watermarks to Text, Image Outputs > Source: > Published: 2026-08-13 16:56:49+00:00 Anthropic will begin watermarking image and text outputs generated by [Claude](/ai/158931/claude) models to comply with the transparency requirements of the European Union's [AI Act](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai#1720699867912-1). The watermarking requirement will apply to Claude models released on or after Aug. 2, the day the AI Act went into effect, and Anthropic will be using two different machine-readable markings for text and image outputs. For text, Anthropic will embed an “imperceptible watermark” directly into the output. You won’t be able to see the watermark, but it will travel with the text and persist through copy-pastes and even some minor edits. “Watermarking will be applied at the model level, which means it will be present no matter which Claude product or surface the text comes from,” the AI startup [says](https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content). For image outputs, including .svg, .png, or .jpg files, Anthropic will attach signed provenance metadata that follows the [C2PA](https://c2pa.org/) open standard already adopted by [OpenAI](/ai/150789/openai-adds-labels-to-ai-images-but-says-metadata-can-be-easily-removed) and [Meta](https://about.fb.com/news/2024/02/labeling-ai-generated-images-on-facebook-instagram-and-threads/). Google also accepts the standard, but uses its proprietary [SynthID watermark](https://deepmind.google/models/synthid/) to label images. Its [Gemini](https://au.pcmag.com/ai/101167/google-gemini) chatbot and a dedicated SynthID detector portal can identify [images generated](/ai/162154/gemini-can-now-detect-ai-generated-videos-but-with-a-key-limitation) by the company’s own AI tools. Anthropic said it is “working to enable users and other third parties to detect Claude’s embedded watermarks and provenance metadata,” but added that the watermarks aren’t definitive evidence for outputs generated by Claude. And while this is intended to comply with EU regulations, Anthropic says its watermarks will apply to outputs generated by Claude models—including Claude Platform (API), Claude, Claude Code, Claude Cowork, and Claude Tag—anywhere in the world. The announcement has made some Claude users, who use the AI tool for writing or photo-editing, [upset](https://techcrunch.com/2026/08/12/some-claude-users-are-mad-that-anthropics-new-watermarks-will-catch-them-cheating-at-their-jobs-classes/), but a large section believes they can [overcome](https://www.reddit.com/r/technology/comments/1vmtw7h/some_claude_users_are_mad_that_anthropics_new/) the detection by paraphrasing the text or using other AI tools to remove the watermark.