Veo 3: A Content Creator’s Playground for…

At first glance, something seems amiss about the body lying in the street. The white sheet covering it is suspiciously pristine, and the police officers’ actions seem entirely aimless. “We need to clear the street,” one officer gestures emphatically, although her lips remain still. It’s clearly AI at work. But here’s the twist: my prompt didn’t ask for any dialogue.

This line was added by Veo 3, Google’s latest AI video generation model. Within 24 hours, I had created a dozen clips depicting various scenarios, from news broadcasts to disasters and even cartoon cats, with audio that was often auto-generated by the model itself. It’s undeniably unsettling and far more advanced than I anticipated, though I doubt it will lead us directly to a misinformation apocalypse just yet. Even so, Veo 3 has all the hallmarks of an AI content generation machine.

During this week’s I/O event, Google unveiled Veo 3, showcasing its groundbreaking ability to produce synchronized sound for AI-generated videos. “We’re entering a new era of creation,” asserted Josh Woodward, Google’s VP of Gemini, labeling it “incredibly realistic.” Initially, I was skeptical, but after prompting Veo 3 to create a video of a news anchor reporting a fire at the Space Needle, I found it convincingly realistic within minutes — all thanks to a simple text prompt and an expensive subscription to Google’s AI Ultra plan.

I decided to try the news anchor prompt after seeing the impressive results generated by Alejandra Caraballo, a clinical instructor at Harvard Law School’s Cyberlaw Clinic. One of her clips showcases a news anchor announcing the fabricated death of U.S. Secretary of Defense Pete Hegseth—who is alive and well. A post featuring multiple AI-generated videos where characters protest their own creation prompts has garnered 50,000 upvotes on Reddit. These videos depict various distressing scenarios, from natural disasters to hospital scenes, all with convincing audio and sound effects. Pretty grim, to say the least!

Perhaps I’m being overly optimistic, but after experimenting with Veo 3, my initial concerns have somewhat alleviated. For one, clear guardrails are established. Prompts for videos showing Biden tripping, an anchor announcing the president’s assassination, or even an exaggerated tech CEO counting cash amidst dollar bills are simply not allowed. That’s a promising first step.

Nevertheless, it can still produce some concerning content. Without needing to get creative, I prompted Veo 3 to animate a video of the Space Needle engulfed in flames. I used a personal photo of Mount Rainier to create a video showing it erupting with plumes of smoke and lava. When paired with a