What Is a SynthID Watermark? Images, Text, Audio & Video
SynthID is Google DeepMind’s invisible watermarking technology for AI-generated images, text, audio, and video. Detection is provider-specific — AI Watermark Center does not detect SynthID.
- Published
- Last updated
SynthID is a watermarking technology from Google DeepMind designed to mark AI-generated content so it can later be identified. It covers images, video, audio, and text in DeepMind’s documentation. It is not a single public detector that every company shares, and it is not the same thing as C2PA Content Credentials. AI Watermark Center does not independently detect SynthID — verification stays with provider tools where they exist.
Modalities
Google DeepMind documents SynthID watermarks in images, video, audio, and text. Image and video watermarks are described as invisible signals added when content is created. Audio watermarks are described as inaudible. Text watermarking adjusts the choice among plausible next tokens rather than inserting hidden characters.
Imperceptible signals
The documented design goal is that people should not notice the mark, and that quality should not be obviously degraded. That does not mean the mark is unbreakable, or that a third party can read it without the matching detector.
Detection differs by media
Gemini Apps currently offer user-facing verification for supported images, video, and audio. Google DeepMind also documents identification of Gemini-generated text. Those surfaces are not interchangeable. AI Watermark Center’s Unicode scanner does not detect SynthID, and the image inspector’s C2PA check is not a SynthID result.
Provider examples
Google documents SynthID across Gemini text and media, including Nano Banana image outputs — see the Gemini SynthID page. OpenAI documents SynthID on supported generated images and audio — see the OpenAI SynthID page. Both provider pages, the database, and the compare view own product behaviour; this guide stays technology-level.
Public detection availability differs
- Gemini’s verifier currently identifies Google AI SynthID, not every company that may adopt SynthID.
- OpenAI publishes a verifier and API for supported OpenAI image and audio signals.
- Claude’s statistical text detector is announced, not integrated here.
- AI Watermark Center does not detect SynthID.
Related pages
- Gemini SynthID — Google product documentation and verification surfaces.
- OpenAI SynthID — image and audio scope OpenAI currently documents.
- C2PA vs SynthID — metadata provenance versus invisible watermarks.
- AI watermark detector — what can actually be checked locally.
- AI Watermark Database — dated provider claim records.
Sources
SynthID — Google DeepMind
Accessed August 15, 2026.
Verify AI-generated images, videos, and audio — Gemini Apps Help
Accessed August 15, 2026.
Provenance signals (Content Credentials, SynthID) in OpenAI-generated content — OpenAI Help Center
Accessed August 17, 2026.
Related pages
- Image Metadata & C2PA InspectorInspect EXIF, privacy metadata and embedded Content Credentials directly in your browser.
- AI Text Watermark ScannerHidden Unicode and invisible character checker for pasted text — zero-width marks, unusual whitespace, and normalization differences. Runs locally in the browser.
- GeminiGoogle DeepMind documents SynthID across image, video, audio, and Gemini text. Gemini Apps add invisible SynthID and Content Credentials on generated or edited media. Gemini’s verifier identifies Google AI SynthID, not every company’s watermark.
- OpenAIOpenAI documents layered image provenance (C2PA and SynthID) on supported ChatGPT, Codex, and API outputs, SynthID on supported generated audio, a public verifier, and a Content Provenance API. A current OpenAI text watermark is not established. Research profile and image cluster available; AWC inspects embedded C2PA only.
- ClaudeAnthropic documents a statistical text watermark for supported new models and C2PA Content Credentials on supported generated files. AI Watermark Center can inspect Unicode and embedded credentials; it cannot currently detect Claude’s keyed text watermark.
- Adobe FireflyAdobe documents automatic Content Credentials on fully Firefly-generated assets, with Adobe as issuer and optional cloud storage for recovery. SynthID use is not established from the documentation reviewed here.
- Gemini SynthIDHow and where Gemini/Google products currently use SynthID.
- OpenAI SynthIDWhere OpenAI currently documents SynthID on images and audio.
- C2PA vs SynthIDMetadata Content Credentials versus invisible watermarks.
- AI watermark detectorWhat can actually be checked locally — SynthID stays provider-side.
- AI Watermark DatabaseProvider records with claim status, evidence, and inspection support stored separately.
- Compare systemsDocumented states across Claude, OpenAI, Gemini, and Adobe Firefly.
- MethodologyHow this site labels documented, verified, observed, and unverified claims.