Marks and provenance

A statistical watermark that tilts word choice while the model writes. Applied to text in the Gemini app since May 2024. The method was published in Nature in October 2024 and released as an open-source library with Hugging Face, so anyone can apply and detect a mark with their own key.

Detection needs the key that made the mark, so a Gemini mark can only be checked by Google or by a party Google authorises.

Reference implementation of the text watermark for Hugging Face Transformers. Useful for anyone who wants to watermark their own model's output, or to study how the signal behaves after editing.

Developer guide to applying SynthID to Gemma and Gemini outputs, with the configuration options and the trade-offs between mark strength and output quality.

Announced August 14, 2026 and updated September 1, 2026. A version of SynthID-Text applied globally to future Claude models, with models released before August 2, 2026 to follow over the coming months. Detection uses Anthropic's key and answers only whether Claude likely wrote part of a passage. Access is open by application to regulators, law enforcement, media, fact-checkers, researchers, educators, EU civil society groups and enterprises with EU compliance duties.

Anthropic's own words: light editing probably will not remove the mark completely; a complete rewrite where every word is replaced will.

Announced October 5, 2026. OpenAI reports that at a 1 percent false positive rate its detector found the mark in about 95 percent of 400-token passages and about 80 percent of 200-token passages, with lower rates on constrained text such as mathematics. Detector access is limited to approved researchers and expert organisations. OpenAI says it plans to open-source the technique.

In OpenAI's own evaluation, replacing 10 percent of words with synonyms cut detection on 400-token passages from about 92 to 66 percent, and replacing 25 percent cut it to 17 percent.

A signed manifest that travels with a media file and records where it came from and what was done to it. Widely adopted for images, audio and video by camera makers, creative software and the large model providers. The EU Code of Practice treats secured metadata like this as one of its two marking layers.

Coverage for plain text is thin, because text is usually copied without any file container to carry a manifest.

Drop in an image, video or audio file and read any attached C2PA manifest, including the signing organisation and recorded edits.