Anthropic’s invisible Claude watermark — can it survive the edit test?
anthropic will watermark all text from claude models per eu rules. the watermark persists through copy-pastes, potentially editing, affecting how ai content is tracked across platforms.
anthropic enabled watermarking for all claude models and files generated after august 2 to adhere to the eu ai act transparency code. the technology embeds identifiers at the model level, ensuring watermarks travel with text during copy-pasting and potentially through editing. support extends to older models and varies products like the claude api and claude code. this implementation signals a shift where compliance drives technical architecture, joining similar recent commitments from google, openai, and meta.