The video generated by AI has gotten too good. In fact, it's scary. That's why our feed has earned millions of views of tiktoks, shorts and reels, like incredibly cute animals bouncing off the trampoline.
It's not always easy to know what's real, as AI content blends seamlessly into scrolling. So how can you tell if a virus video is generated by AI?
The truth is, there is no perfect checklist for finding videos generated in AI. Negar Kamali, an AI research scientist at Northwestern University's Kellogg School of Management, told Mashable Tech Reporter Cecily Mauran last year.
How to identify video generated by AI
Old prizes – distorted faces, broken fingers, incredibly smooth textures – are becoming more difficult to catch as the technology improves. Temporary inconsistencies have been cleaned up. But just like the surreal animal clips filmed on a fake doorbell cam, the truth still lives in small details. That's where the synthetic mask always slips.
The rise of surreal AI video tools
Part of the challenge is the technology itself. Tools like Openai's Sora and Google Veo 3 can now generate film clips with complex camera movements, realistic lighting and incredible textures. These platforms aren't just for toys. It is framed in the realm of professional-grade filmmaking, and the gap between human shot footage and AI-generated content is thinner than ever before. This means that when you discover “Tells” in a virus AI video, you will sharpen your eyes and become a little more skeptical.
For example, consider a video of an entire bunny where the entire workshop bounces back in perfect rhythm. It's absolutely adorable (and easy to make), but looking a little closer, it's deeply suspicious.
With that in mind, here's the best way to identify virus videos generated by AI:
1. Look at the context first
Many AI videos are staged in strangely specific scenarios. Most often, I use Onyx-Filter Night Vision. It's not just about “aesthetics.”
Dark filters conveniently hide small glitches and frame contradictions that are common in AI footage.
Masculine light speed
2. Check the missing device features
If you claim that the video comes from a doorbell cam or security feed, look for timestamps, brand logos and interface overlays. The complete absence of these is doubtful. At the same time, the presence of these features does not necessarily mean that the video is authentic.
3. Look at physics
Real-world movements follow real-world rules. For example, animals should not perform perfectly timing repeat jumps straight for 10 seconds. Take a look at the tip of this whale, for example.
4. Be careful of the runtime
Shorter clips will have fewer opportunities for AI to reveal its flaws. So, just before something appears to be “off”, a huge number of synthetic videos of the virus are blocked.
“If the video is 10 seconds long, it's suspicious. That's why it's so short,” Honey Farid, a computer science and digital forensics expert in Berkeley, California, told Mashable.
Similarly, longer videos are suspicious if very short clips are sewn together. Most AI video generators can generate only short clips. The most advanced generation AI video model, Google Veo 3, generates 8-second clips. Sora by Chatgpt-Maker Openai creates 1-20 second videos.
5. Listen to the sound (or lack of it)
Composite clips often have oddly clean audio, ambient noise mismatches, or none at all. “It comes out of them and distorts certain facts…it's really hard to disprove,” Aruna Sankaranarayanan, research assistant at MIT's Institute of Computer Science and Artificial Intelligence, told Mashable. A silent or overly clean soundscape can be a big clue.
6. Spot AI-Text Artifact
AI is still struggling with easy-to-read writing. Check out the clothing, signs, or packaging in the frame – distorted letters, random symbols, or meaningful text are permanent prizes. “If the image feels like ClickBait, it's ClickBait,” Farid told Mashable.
For example, this viral video of an emotional supportive kangaroo. Take a closer look at its best when the video zooms in.
7. Beware of impossible movements
Humans and animals have subtle weight shifts, irregular walking patterns, and micro-animal animals. AI works often lack these subtleties. And if you look closely, you can find strange contradictions such as multiple numbers melting into one or vice versa.
“The building has either added a story or the car has changed colour. It's physically impossible,” Farid said Mashable, explaining the temporal contradiction.
8. Find the watermark (or be careful not to be present)
Some AI video generators, including SORA and VEO 3, automatically embed watermarks or metadata to identify synthetic content. These marks can appear in the corners as faint overlays or as hidden digital signatures within the file. Digital watermarks like Google Deepmind's SynthID are promising, but watermarks can also be removed or cropped from the viral video.
9. Check your account history
Many AI videos will be eliminated Mars AI Slop Farmers. If you see a video that appears to be out of place, check the account behind the video. Often, you can see that they have posted dozens (or hundreds) of nearly identical AI videos in a short period of time. It's a big red flag that the video you just saw was generated by AI.
topic
Artificial Intelligence Social Media
