Technical overview
How Claude watermarks work
Anthropic describes two complementary marking layers: an imperceptible watermark embedded by supported models in text, and signed C2PA provenance metadata attached to supported files. They answer different questions and fail in different ways.
Evidence boundary: Anthropic says third-party detection details are forthcoming. ClaudeWatermarks does not present ordinary Unicode or clipboard residue as Anthropic's model-level watermark.
Layer 1: embedded text watermarks
Anthropic says the mark is woven into generated text at the model level, travels when text is copied, and may survive some editing. Its current public article does not disclose the encoding or publish a general third-party detector, so ordinary character inspection cannot be equated with official detection.
Layer 2: signed file provenance
Supported files can carry signed metadata following the C2PA open standard. A verifier can read the manifest, validate hashes and signatures, and distinguish a valid credential from a file that was altered after signing. Re-saving, converting, or taking a screenshot can strip metadata.
Why the results are signals, not authorship proof
A valid mark may show that Claude processed content without proving Claude originated every idea or word. Conversely, no detected mark may reflect unsupported models, editing, short text, stripped metadata, or an unsupported platform.