What are Google Veo 3 SynthID watermarks?

Google Veo 3 SynthID watermarks are invisible signals that DeepMind's SynthID tool embeds into AI-generated content at the moment a Google model creates it. They cover images, audio, video and text, and are readable by SynthID's technology but not visible to humans.

How do I check if content has a SynthID watermark?

Upload an image, video or audio clip into Gemini and ask if it was made or edited by Google AI. Gemini checks for the watermark and reports whether one is present. A separate SynthID Detector portal is also in early testing with journalists.

Can SynthID detect all AI-generated content?

No. SynthID detects Google's own watermark only, not content from every AI tool. It is not a universal lie detector. Google frames it inside a wider provenance effort alongside OpenAI's work and the C2PA Content Credentials standard, which aim for a shared, interoperable trust layer.

How does SynthID watermark text without changing it?

Large language models pick each word using probability scores. SynthID nudges those scores in a controlled pattern to leave a hidden signature. This marks the text without altering its meaning or quality, so readers see no difference in the output.

Does the SynthID watermark survive editing?

Google says the audio watermark survives noise, MP3 compression and speed changes. For images and video, it is designed to survive cropping, filters, frame-rate changes and lossy compression. The mark is built to remain readable after common edits, though it detects only Google-made content.

All insights

AI Search

4 min read28 August 2026Nathan Mzumara

Google Veo 3 SynthID Watermarks: How Google Marks AI Video, Text, Images and Audio

Google Veo 3 SynthID watermarks are invisible signals that Google embeds into AI-generated content so software can later tell it was made by a Google AI model. SynthID, DeepMind's watermarking tool, now marks AI-generated images, audio, video and text across Google's consumer products. Gemini can read those signals when you upload a file and ask.

If you build a brand on content, this is the moment provenance stopped being a policy slide and became a technical fact. In my opinion, this is one of the more consequential shifts of the year for anyone in search, discovery or marketing. It changes how audiences, platforms and regulators decide what is real.

What are SynthID watermarks?

SynthID is a DeepMind tool that embeds a digital watermark directly into content the moment a Google generative model creates it. The watermark is invisible to humans but readable by SynthID's own technology. It now covers all four content types: image, audio, video and, most recently, text.

Text is the interesting one. As Google explains, large language models generate one word, or token, at a time, and each candidate word carries a probability score. In the sentence "My favourite tropical fruits are mango and...", the word "bananas" scores higher than "airplanes". SynthID nudges those scores in a controlled pattern to leave a signature. It does this without changing the meaning or quality of the output.

Google has also shipped a way to check for the mark. You upload an image, video or audio clip into Gemini and ask if it was made or edited by Google AI. Gemini reports back if it finds a SynthID watermark. There is also a separate SynthID Detector verification portal, currently in early testing with journalists and media professionals.

When did Google roll this out, and how does it work?

Google's image, audio and video watermarking has been live across its consumer AI products for some time. The text watermarking is the newer expansion, covering the Gemini app and web experience. The SynthID Detector portal launched earlier this year and is still gated behind an early tester waitlist.

Here is the mechanism in plain steps:

You generate content with a Google model such as Gemini, the Lyria music model, or Notebook LM's podcast feature.
SynthID embeds a watermark at the point of creation. For audio, Google says the mark survives noise, MP3 compression and speed changes. For images and video, it is designed to survive cropping, filters, frame-rate changes and lossy compression.
To verify, you upload the file to Gemini, or the Detector portal, and ask if it was AI-generated or edited.
Gemini checks for the watermark and tells you whether one is present.

There is one important caveat. SynthID detects Google's own watermark. It is not a universal lie detector for all AI content. That is why Google frames this inside a wider provenance effort, and why OpenAI's work on content provenance and the C2PA Content Credentials standard matter here too. No single vendor owns the truth. The direction of travel is a shared, interoperable trust layer, not one company's badge.

Why do SynthID watermarks matter for growth and discovery teams?

From what I can see, three things change the moment content becomes verifiable at scale.

First, trust becomes a ranking-adjacent signal. Google has been building toward identifying AI-generated media online for a while. Once platforms can reliably label content, provenance moves from nice-to-have to expected. Brands that can prove how their content was made will earn a quiet advantage in discovery.

Second, your own workflow is now auditable by others. If your agency ships a "hand-crafted" asset that carries a SynthID watermark, a journalist or a client can check it. This could expose gaps between what you claim and what you actually did.

Third, verification becomes a customer-facing feature. Media, publishing and any regulated sector can now offer proof of origin as part of the product, not a footnote.

What about content you did not make with Google?

Most of the internet is not watermarked. That is a fair objection. The answer is to build your own provenance trail now rather than wait for universal coverage. This links to the move toward watermarked text across the major model providers. It also sits inside the wider question of who governs and edits AI-produced work before it reaches an audience.

The concrete action to take this quarter

Do three things this quarter. First, audit your content pipeline and log which assets are AI-generated and with which tool, so a watermark check never surprises you. Second, adopt C2PA Content Credentials where your tools support them, so your provenance is portable and not tied to one vendor. Third, write a simple internal disclosure standard now, before a client or regulator asks for one.

What Google Veo 3 SynthID watermarks mean for what you publish

The era of "you can't tell" is closing. Google Veo 3 SynthID watermarks turn content origin into something anyone can verify with a single upload. In my view, the teams that treat provenance as a feature, not a liability, will be the ones people trust to publish.