AI video made with Google's new Veo 3 leaves viewers 'genuinely terrified' at how realistic the footage is
They say you can't always believe everything you see online, and with a new era of AI video generation coming into play, that's more evident than ever.
There's been a massive boom in this kind of software, and while it was once a clunky mess that couldn't generate people's fingers properly, you might now be unable to tell the difference between what's the real deal and what's been made by AI.
The likes of Sora and Runway are muscling in alongside the big boys, and keen not to be left behind in the AI wars, Google's own offering is coming out swinging.
Veo 3 is an AI generator that incorporates audio and video to compete with OpenAI's Sora. Eli Collins, Google DeepMind Product Vice President, wrote in a blog how it will include the likes of dialogue between characters and animal sounds: "Veo 3 excels from text and image prompting to real-world physics and accurate lip syncing."
Advert
Veo 3 is now available to US users who are part of Google's $249.99 a month Ultra subscription plan. It's also available for users of Google’s Vertex AI enterprise platform, although both are geared to hardcore AI enthusiasts rather than casual users.
Over on YouTube, an impressive video from László Gaál tests Veo 3 and plays out as a non-existent car show. Some of the dialogue is a bit off, like one guy saying, "Man, the acceleration is crazy, you look far, step on the pedal, and you are there."
Impressively, there's a wide array of different nationalities, voice styles, age ranges, and body types, teasing what Veo 3 can really do. If you look at the background characters, some are slightly clumsy, but if someone hadn't told you the video was AI, you likely wouldn't know.
Others in the comments were equally impressed, with one concerned viewer writing: "This is both impressive and scary all at the same time."
Another added: "Even though the technology to create this is amazing. I am also worried. At some point whatever you see through a screen might be fake in the future. It is amazing but also scary at the same time.”
A third said: "The implications here are absolutely terrifying, there is no turning back now."
Saying the media is 'absolutely cooked', it does raise questions about what happens next. We've already seen some pretty realistic deep fakes, like when someone hijacked government screens with a video of Donald Trump sucking Elon Musk's toes, but they look amateur when compared to what Veo 3 could do.
It's not just Veo 3 coming hot out of the gate, with Google also announcing Imagen 4, which is able to produce higher-quality images through user prompts. Finally, there's Flow, a filmmaking tool that can create cinematic videos simply by describing locations and style preferences. Google's AI future is here, and with it, it's clear many of you are worried about what will happen next.