Google had an amazing I/O 2025 event a few weeks ago, stunning the world with several product announcements. It wasn’t just about showing off impressive Project Astra features that should be available in Gemini Live soon. Google also released several products that are already available to premium Gemini subscribers.
The Veo 3 AI video generation tool is one of the most impressive products from I/O 2025. The video generator lets you add sound to your clips, including background noise, special effects, and speech.
We’ve seen plenty of videos generated with Veo 3 since its release that make you question reality. They look so real it’s almost impossible to tell they were created with AI. One of the best examples is a short movie that The Wall Street Journal created using AI tools, including Veo 3.
All AI-generated content made with Google tools is invisibly watermarked with SynthID. That’s a problem for most people, including fans of AI products. I know all Veo 3 clips are watermarked with SynthID, but there’s no easy tool available online to verify that.
Now imagine the average internet user exposed to an increasing avalanche of content. They won’t know how to use SynthID and likely won’t care while doomscrolling. That’s the target audience for anyone looking to abuse tools like Veo 3 to create misleading clips that go viral online.
Luckily, Google has decided to add a visible watermark to most videos created with Veo 3 as it expands access to more Gemini subscribers globally.
Google expanded Veo 3 access to the UK on Friday, while the Gemini AI Ultra plan rolled out to 73 countries. The EU is still missing from that rollout, so most users in Europe can’t subscribe to Google’s $249.99/month plan, which gives you the highest limits for Veo 3 and other advanced AI models.
Google AI Pro subscribers in these countries can access Veo 3 in Flow. They also get a 10-pack of trial video generations in the Gemini apps, including on mobile.
The visible watermark detail is buried in the blog post announcing the Veo 3 expansion. With more people now able to use a tool that creates videos resembling footage from a professional camera or a phone, Google likely realized that a visible watermark was necessary.
All videos made with Veo 3 will carry a visible watermark, except those created using Veo 3 in Flow by Ultra subscribers. As seen in the video above and in the screenshots from the same clip, the Veo 3 watermark appears in the lower right corner. It isn’t visible in every scene, and the font could be bigger. But it’s something. It’s a start that can help Google curb Veo 3 abuse.
Whether or not the visible watermark is present, all Veo 3 videos, including those made with Flow, will continue to carry the SynthID watermark.
The clip at the end of this post briefly explains SynthID. It’s a way for Google to watermark AI images, videos, audio, and text. The issue, again, is that it’s invisible. Google treats that as a feature, but I disagree. Verifying the authenticity of content, especially videos created with Veo 3, should be simple for anyone.
Invisible watermarks mean someone has to do extra work to figure out where a video came from. Google says its SynthID Detector tool can help, but it isn’t available to users. Like I said earlier, I can’t just pick a Veo 3 clip, drop a link into a Google tool, and get a result.
Google rushed to make Veo 3 available to the public before rolling out its SynthID Detector widely.
That said, it’s still good to see Google taking action now. If Veo 3 is already this good at generating lifelike videos, imagine what Veo 4 and future versions will be capable of. By then, more people will be used to seeing Veo 3 watermarks and using the SynthID Detector Google is currently testing.
To see how SynthID can detect AI-generated content, visit Google’s pages at this link.
