- YouTube Shorts now offers the Veo 3 AI video generator for producing 8-second videos from text prompts
- Veo 3 will make cinematic or animated clips with customized motion, lighting, and native audio
- Every AI-generated video is marked with a SynthID watermark and a visible AI label
Google has embedded its most advanced AI video generator, Veo 3, into YouTube Shorts.
You can write simple text prompts inside the YouTube mobile app and have the AI make a video clip, including sound, that lasts for up to eight seconds.
You type something like “a hummingbird flying through a neon jungle at sunset” or “a robot chef flipping pancakes on Mars,” choose a visual style, like a kind of animated or cinematic look, and Veo 3 will quickly render your dream mini-movie.
The streamlined “fast” version of Veo 3 may be limited in video length and quality, but the Shorts-native version is still impressive considering it’s produced on a phone with no actual film footage involved.
This integration builds on Veo’s official debut earlier this year and expands on the Dream Screen feature that lets users generate AI backgrounds for Shorts. But Veo 3 ups the ante with more than just a backdrop. It produces full standalone videos.
Google boasts that Veo 3 is good enough to make natural-looking camera movements, while tracking multiple objects and offering stylistic consistency.
That includes how objects move in space. Submit a prompt like, “a knight riding a horse through a burning village”, and you’ll get a choreographed action scene, not just the disparate elements mashed together. The horse’s movement, the knight’s arms, and the flickering flames move independently and in reaction to each other.
You can play with Veo 3 from the YouTube app by tapping Create and choosing Create a video from a prompt. You don’t need to upload anything or have a Gemini subscription at the moment.
Watch On
There’s even sound, though no full dialogue as of yet. Still, you can get atmospheric and thematic music generated based on your scene.
Giving a forest scene the sound of wind and birds or hearing the hums and beeps of a spaceship makes the final clips feel more finished than a silent film can manage.
Don’t think you can just trick people into thinking your crazy video is from the real world though. Google added safeguards to prevent that, with every clip generated with Veo 3 in Shorts automatically tagged with a label that says “AI-generated” and embedded with the invisible watermarking of SynthID.
AI visions
Google is betting that for viral Shorts, meme-makers, visual poets, and other untrained users, that won’t be a dealbreaker.
And the content doesn’t have to stay AI-only. The tool also allows users to generate green-screen video backgrounds, which can be layered behind human performers.
You could record yourself in your bedroom and replace the background with a dramatic space opera landscape, all from your phone. The possibilities are potentially as broad as your imagination.
For everyday users, the real question is: what do you do with it? Make surreal memes? Create ambient video loops for music tracks? Build AI-powered reaction backgrounds?
The barrier to entry for video creation is now so low as to be underground. You don’t need gear or editing software. It’s arguably as big a deal as the initial release of a mobile YouTube app with video creation options.
Of course, that’s not necessarily a good thing. YouTube has lots of weird and annoying nonsense produced in some way with AI.
This could flood YouTube Shorts with even more unpleasant weirdness. And flooding the Shorts feed with too much AI content could overwhelm human-made work.
Watermarks help, but they don’t stop people from using generative tools in misleading ways. Nonetheless, Google is more focused on how it will make AI-generated video part of the everyday internet experience, skateboarding penguin and all.
