I don't spend any more time on Facebook than I should, but I'm still exposed to countless AI slop videos in my feed. It's easy to find them considering what I do for a living. But worryingly, many people who flocked to the comments don't seem to understand that they're looking at something that never happened.
As the resident family geek (you're on this site, after all), it's your job to make sure people of all ages understand what these videos are and why they're dangerous at the annual Klan gathering.
Why are fake videos everywhere now?
What do these fake videos mean? The answer is usually engagement. People earn advertising revenue by posting video content that gets a lot of attention. Ads running during fake AI videos bring in the same revenue as ads running during legitimate clips. Some may be politically motivated or simply want attention. The tools for creating this content are inexpensive and easy to use. For example, you may have seen a viral clip on YouTube of a raccoon jumping on a backyard trampoline, but it's completely fake.
Creating videos like this is very easy. to show you How easy was that?We asked ChatGPT to create an image of us jumping on a trampoline with an AI raccoon. All I had to do was provide two sample images and a prompt asking me what I wanted.
They then used Google's Veo 3 AI video generator's “frame to video” feature in Google Flow to display prompts to explain the action.
A few seconds later, I got a video of someone who looked like me (but not quite) jumping on a trampoline with wild animals.
With just a little money and a little time, you can create dozens or hundreds of videos a month. Only a portion of them needs to go viral to be economically viable, so it's no wonder people would rather generate videos than go to the effort of creating them the old-fashioned way.
The most common types of fake videos that people get fooled by
Unfortunately, even before AI became this easy, people were already experts at knowing how to manipulate social media users and get their attention. The trick is to present videos that evoke some kind of strong emotion. “Ragebait” occurs when someone manipulates you into being angry. But other emotions can come into play as well. This is why we see so many fake heartwarming stories about dogs rescuing injured veterans and babies from danger.
Now, if you think about what you see in these clips, you'll see that they're more than just concepts. However, there are clearly many people who think these are real (even considering bot accounts) and who have a slightly distorted sense of reality because of this.
Subtle signs that a video isn't real
Let's be clear: trying to spot AI videos by eye is becoming a fool's game, and it's becoming more so with each passing day. At some point, the AI models that generate these videos will be so good that only specialized tools will be able to detect them, and they may eventually become useless.
It's safer to assume that what you see is generated by an AI, especially if it's illusory or designed to provoke you. However, there are still some tell-tale signs.
-
Strange physics and movement of people and things.
-
Attempts to remove watermarks, such as the moving watermark used by the Sora video generator. This usually appears as a bouncy, blurry spot.
-
Similar to the original raccoon video, some AI videos show objects randomly transforming, merging, appearing, and disappearing.
-
Garbled text on signs and clothing.
However, some real-world videos contain video artifacts, and you don't need to be an AI to direct or manipulate a video. The internet is a wild place, sometimes with little room for truth.
How to talk about it without starting a family quarrel
When bringing up the issue of fake videos with family members, it is very important to avoid making anyone feel stupid. Not everyone follows technology news as closely as we do. Most people have heard of AI, but don't know exactly what the technology is capable of or how quickly it's progressing. So who can blame them for believing a video they saw on social media during a casual doom scroll?
Think of this as a global shift in technology that requires new forms of literacy. As long as users are aware that any videos they encounter online may be fake, it's enough to inoculate them against the worst scams and engagement baits. There's a good chance you don't even need to bring it up. Because as soon as you walk in the door, your uncle or aunt could shove an AI video in your face.
