Video: How to spot AI videos in 2025

Comments Comments

AI videos are becoming increasingly common, and their quality has improved so much that spotting them is getting tougher. 

It's one reason why this video from Jeremy Carrasco is worth watching, as he breaks down a few of the 'tells' that will help you identify the real from the fake. 

Jeremy points out that spotting the bad AI videos is straightforward, but it’s the good ones that slip into our feeds almost unnoticed that are much harder to spot - take this one below as an example.

Would you have automatically assumed it was genuine or AI generated?

There are a bunch of tell-tale signs to identifying AI, and Jeremy puts these into three categories: AnatomyTechnical, and Situational. While ticking one of the boxes won't be a giveaway on its own, stacking a few together is a pretty surefire way to guarantee something is AI-generated.

Anatomy includes things like weird, waxy skin textures (often a big giveaway), odd hand and body movements (AI struggles with actions like running and sports in particular), and waxy, glassy eyes. 

Technical covers more subtle giveaways: that soft, slightly blurry look even in high resolution video, a 'cinematic' 24 fps frame rate in content you wouldn't expect to be captured in this style, imperfect lip sync or robotic voices, and strange background glitches or odd object morphing.

And finally, Situational covers the material itself. Does it feel too slickly produced for what is 'throw away' content? Does the content play on emotions and seem ungenuine? Is it trying to go viral?
 
Although, sadly, we are now pretty much conditioned to slop content of this type clogging our feeds, understanding why its been put together can help you decide whether its authentic or not. 
 

The reality of all this is that AI video is getting better and better, and it is becoming increasingly hard to identify. You owe it to yourself to stay wary and curious about what you're seeing. 

comments powered by Disqus