SynthID uses AI to digitally watermark any generated image, and Google generative technologies such as Veo carry this watermarking.
Resolved true The record shows it happened or was so.
Factual: Statements about the present or the past, checked against the record: did they hold up?
The second judge would not resolve this claim; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Test, as of 10 April 2026, on two parts: (a) SynthID uses AI (a machine-learning model) to embed a digital watermark in generated images, and the claim's 'any generated image' is read as images produced by Google's own generative tools, not images from every generator; (b) Google generative technologies such as Veo carry SynthID watermarking. True: Google documentation or technical publications show that SynthID uses a learned or AI-based method to embed imperceptible watermarks in images generated by Google's tools, such as Imagen and Gemini image generation, and that Veo outputs carry SynthID. Mostly true: both parts hold, with a minor slip, for example certain Google image outputs or surfaces not being watermarked, or the 'uses AI' description being loosely accurate. Mixed: only one of the two parts holds. Mostly false: SynthID watermarks only a narrow subset of Google-generated images, or Veo is not watermarked, or the claim implies that SynthID watermarks images from any generator, including non-Google ones. False: SynthID does not watermark generated images, or Google's generative tools such as Veo do not carry it.
Set before the research began.
Evidence (4)
-
E1 Identifying AI-generated images with SynthID Google DeepMind · 29 August 2023 Own record
“SynthID uses two deep learning models — for watermarking and identifying — that have been trained together on a diverse set of images.”
-
E2 SynthID-Image: Image watermarking at internet scale arXiv Own record
“SynthID-Image has been used to watermark over ten billion images and video frames across Google’s services”
-
E3 A developer's guide to Imagen 3 on Vertex AI | Google Cloud Blog Google Cloud · 30 August 2024 Own record
“Imagen 3 utilizes SynthID, a technology which embeds an invisible watermark at the pixel level. By default, a digital watermark is added to all Imagen 3 generated images”
-
E4 fal-ai/veo3.1 fal.ai · 4 November 2026
“All videos generated with Veo 3.1 are invisibly watermarked with SynthID, Google's AI content detection technology.”
Reasoning
Google DeepMind's launch post (E1) says SynthID uses trained deep learning models for watermarking. That supports the claim's "uses AI" description. Google's technical paper (E2) says SynthID-Image has watermarked over ten billion images and video frames across Google's services, and Google Cloud's guide (E3) says all Imagen 3 images get a SynthID watermark by default. A third-party model page (E4) says all Veo 3.1 videos are invisibly watermarked with SynthID. E4 was captured after the judging date, but its statement is consistent with E2's mention of watermarked video frames. Read as covering Google's own generative tools, both parts of the claim are supported.
Cite this claim
Demis Hassabis, “SynthID uses AI to digitally watermark any generated image, and Google generative technologies such as Veo carry this watermarking.” (said 10 April 2026). Verdict: resolved true. Reliability API, claim c_ee0ab3b0a28a, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_ee0ab3b0a28a
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Demis Hassabis
-
Mostly true
“Performs better than 3.1 Pro on coding & agentic tasks”
Gemini 3.5 Flash performs better than Gemini 3.1 Pro on coding and agentic tasks.
-
Mostly true
“It can take photos, video & audio and build entirely new scenes.”
Gemini Omni can take photos, video and audio and build entirely new scenes.
-
Mostly true
“I didn’t feel like that was proper AI because that system, let’s take Deep Blue, okay, it’s world champion level at chess, at chess, but it can’t do anything else.”
Deep Blue was world-champion level at chess but could not do anything else.
-
Mostly true
“They still can't really yet go beyond asking a new novel question or a new novel conjecture or coming up with a new hypothesis that has not been thought of before.”
-
Mostly true
“they certainly have modeled enough of the dynamics, put it that way, that they can pretty accurately generate whatever it is, eight seconds of consistent video”
Verdicts are made automatically; how verdicts are made.