Anthropic is embedding invisible, machine-readable watermarks into content generated by Claude, including text, images and files, the company confirmed. The system began rolling out with models released on or after 2 August 2026, as part of efforts to meet transparency requirements under the European Union’s AI Act, a media blog on the Claude support portal stated.
Although the measure is linked to European regulations, Anthropic says the watermarking applies globally across Claude products, including the Claude Platform, Claude Code and cloud services such as AWS and Google Cloud.
“What our marking commitments mean for Claude:
- New models will mark AI-generated content from day one. Claude models launched in the EU on or after August 2, 2026 will support machine-readable marking at launch. Generated text will carry embedded watermarks, and generated files will include digitally signed provenance metadata where supported.
- Marking works everywhere you use Claude. Marks will apply to output from supported Claude models across Claude Platform (API), Claude, Claude Code, Claude Cowork, and Claude Tag, and wherever Claude is offered, worldwide. Some platforms or features may not support certain marking types.
- We’ll help you detect Claude’s marks. We’ll support users and other third parties to detect Claude’s marks, as the Code requires, and we’ll share details in forthcoming documentation.
- Existing models are in progress. The law includes a transition period for Anthropic models launched before August 2, 2026, and we’re working to add marking support for those models as well.”
For text, Anthropic uses a statistical signature incorporated into the model’s word-selection process. The watermark can remain detectable when text is copied and pasted and may survive light editing, although extensive rewriting can weaken it.
Images and files use C2PA metadata, a cryptographically signed provenance standard designed to identify where digital content originated. However, this metadata can be removed through actions such as taking screenshots, changing file formats or uploading content to some social media platforms.
The system, though, has limitations, Anthropic has acknowledged. Short passages may not contain enough information for reliable detection, while text originally written by humans could still receive a watermark after being proofread, translated or reformatted by Claude.
Users cannot disable the watermarking system.
The move has sparked criticism among writers, developers and other Claude users, particularly over concerns that AI-assisted editing could cause human-created work to be incorrectly identified as AI-generated. Some users have also expressed frustration over the lack of an opt-out option and suggested switching to alternative AI systems.
Anthropic claims the rollout marks another step toward greater transparency around AI-generated content, but the debate over watermark reliability, privacy and user control is likely to continue as regulators and technology companies develop new standards for identifying synthetic media.
