You’re Not Ready for This Terrifying New Wave of AI-Generated Videos
I don’t want to sound alarmist, but I think it’s time to start assuming that everything you see on the internet is fake.
The internet is certainly full of content created by real people (including this article). But AI-generated media is becoming so realistic that it almost puts you at a disadvantage when you assume that the content you scroll through your feeds is genuine.
Don’t skip this article because you know what AI content looks like — the current stuff your algorithm is delivering to your social media feeds is easy to spot if you know what you’re looking for. But even if you can spot AI junk the second it hits your eyes, you need to know that you’re not ready for the next wave of AI-generated videos. That wave isn’t just coming — it’s already here.
AI content is already misleading people
Most of us are acutely aware of the kind of “AI video” we see: That “tragic” video of a cat owner saving his kitten by throwing it out of a burning plane is obvious AI junk to most who watch it. You probably know that Trump doesn’t work at that construction site , and you can probably tell that this family of cat farmers is actually AI-generated.
But there are videos that aren’t so obvious, especially to those of us who aren’t as familiar with AI or technology in general. You might know that that video of babies dancing in a circle is AI, but many people in the comments don’t know that (unless they’re bots). You might also guess that that family of pets isn’t actually watching a bird examine a toy alligator, but then again, many people can’t. And there’s no end to America’s Got Talent videos that feature “realistic” but impossible visuals — which still capture the hearts of hundreds of thousands, if not millions, of people. (I’m crying.)
But I’m not writing this article today because I’m worried about how many of these “believable” AI videos are tricking too many people into thinking they’re real. I’m worried about that, but those worries pale in comparison to my new fears.
So far, most of the AI videos taking over social media feeds have relied heavily on visuals and background sounds to sell their supposed authenticity. You’ll notice that none of the characters in any of these videos actually speak. If they do, it’s immediately off-putting, with out-of-sync lip movements and typically robotic voices. It’s been easier for AI creators to emphasize the realism of the people and animals in their videos and hope that you’re sufficiently smitten by the baby-lion dance to not think, “This is bullshit, right?”
Even OpenAI’s Sora video model, which blew me away with its quality last February , relied on realistic visuals. A video of a woman “filming” her reflection through a train window is too real for comfort, but Sora didn’t spit out fully rendered conversations. If you see a scene like that in your feeds, you’d probably assume, of course, that it’s real video — or at least human-generated.
AI-powered video is about to change completely
Something happened this week that only made me more pessimistic about the future of truth on the internet. During Google I/O this week, Google unveiled Veo 3 , its latest AI-powered video model. Like other competing models, Veo 3 can generate very realistic sequences, which Google showed off throughout the presentation. Not great, but nothing really new either.
But Veo 3 isn’t just capable of generating video that can trick your eye into thinking it’s real: Veo 3 can also generate audio to go along with the video. This includes sound effects, but also dialogue — dialogue that’s lip-synced.
To demonstrate the audio/video capabilities of Veo 3, Google showed a clip of an old sailor at sea . The video quality is crisp and realistic, and the words the man says are in sync with his lip movements. Of course, knowing that the video is AI, you notice the oddities that give away the game (to me, it looks more like high-quality animation than live-action footage), but I’m pretty sure this video will fool many fans of fake AGT videos.
But it wasn’t even that clip that triggered my new fears — it was the videos that people started making as soon as they got their hands on the Veo 3. PetaPixel has a great roundup of some of the “best” Veo 3 videos people have made so far, but I’ll highlight a few that should freak you out the most.
This clip shows a streamer playing Fortnite . Everything, including the gameplay footage, was created using Google’s AI:
This tweet is currently unavailable. It may be loading or has been removed.
This clip shows three concerts that never happened, with musicians and audiences that don’t exist. The music isn’t great, but that’s not the point. The music, from the vocals to the instrumentals, was entirely generated by AI, then lip-synced to drums, guitars, and strings:
This tweet is currently unavailable. It may be loading or has been removed.
But this clip is definitely one that should alert all of us. Someone has generated a fake video of a fake car show, complete with fake interviews with fake attendees. It’s far from perfect, but any AI weirdness is completely overshadowed by the superficial realism. Not only would fans of AI’s Got Talent buy this, I would buy this, especially if I wasn’t looking for it:
This tweet is currently unavailable. It may be loading or has been removed.
It’s the visuals; it’s the dialogue; it’s the crowd; it’s the lighting; it’s the genuine laughter at “mistakes”; it’s the sound of a microphone being “bumped into.” Congratulations on noticing that the dialogue often doesn’t make sense, or that the people in the background are defying the laws of physics — you won’t notice it until you’re halfway through scrolling on TikTok or Instagram.
Even the Veo 2, which isn’t as powerful as the Veo 3, now offers tools for realism, like the ability to dictate how you want the camera to move. And both models are available in Flow, Google’s AI-powered video editor of sorts. Creators now have the ability to create highly realistic AI-powered content that feels like it was shot in person, and the technology is only getting better.
This tweet is currently unavailable. It may be loading or has been removed.
Google’s best AI-powered video generation tools cost $250 per month for the new AI Ultra subscription plan . That’s pricey, but it’s affordable for many people interested in creating AI-powered content. But the $20 per month AI Pro plan still includes access to Veo 2 and Flow. The speed limits are lower, but I wouldn’t be surprised to see some realistic goo coming out of those limits.
It’s time to become a real skeptic.
None of this technology is perfect. I’m not here to tell you that everything Veo 3 spits out is indistinguishable from the real thing, or that the videos lack any of the usual AI cues. In fact, there’s something clearly wrong with Veo 3’s training data: As 404 Media reports , the model consistently generates the same weird “dad joke” when you ask for the stand-up comedian’s generation.
What I’m saying is that it’s time to turn on your bullshit detectors and keep them active at all times. When engaging with online video, especially short algorithmic clips, you may be safer operating under the assumption that the content is fake from the start, and demanding proof beyond any reasonable doubt that what you’re seeing wasn’t created with a simple tip and a $250 budget. It seems extreme, but after what I saw this week, I don’t see any other way to engage with this content going forward.
We’re in scary territory right now. Today it’s demos from musicians and streamers. Tomorrow it’s a politician saying something they didn’t say; a suspect committing the crime they’re accused of; a “reporter” feeding you lies via “news.”
I hope this is the best that can come out of this technology. I hope that AI companies run out of training data to improve their models, and that governments take some action to regulate this technology. But given that Republicans in the US passed a bill that included a ten- year ban on government regulation of AI , I’m pretty pessimistic on that last point.
In all likelihood, this technology will continue to improve, without any restrictions to ensure its safe progress. I can only wonder how many politicians who voted yes on this bill watched an AI-generated video on their phone this week and didn’t think much of it.