AI Video Should Be Illegal
Deep Dives
Explore related topics with these Wikipedia articles, rewritten for enjoyable reading:
-
Deepfake
15 min read
The article centers entirely on AI-generated video deepfakes but assumes reader familiarity with the technology. The Wikipedia article covers the technical history, detection methods, legal responses globally, and documented cases of misuse - providing essential context for understanding the policy debate the author is making.
I.
AI video should be illegal.
Let me start my case with a straightforward example I got from astrophysicist Neil deGrasse Tyson’s YouTube channel:
That is scary.
Imagine a deepfake of this quality of Donald Trump saying something barbarous like “We’re gonna do a third term, okay? They don’t want me to, but we’re gonna do it. Everybody’s saying it: Our country is in serious trouble, folks.”
Sounds crazy? Here’s what Nano Banana 2, a new image model by Google, can do:
Within three to six months, you will have the same thing for video.
A “Trump third term” video deepfake doesn’t sound so crazy now, right? I can no longer tell that this image is a fake, just to be clear. I know only because Google has somehow implemented an invisible (and hopefully indelible) watermark (good for them). But the tech is well past the point of obvious telltale cues (where did we leave the crooked fingers and six-finger hands?).
I’ve heard people suggest there’s a reasoning model (Gemini) underlying Nano Banana 2, and that’s why it’s so good. Whatever the case: AI can now make real anything you want to believe and thus anything others want to believe as well.
“Fair, Alberto, but who’s gonna believe a fake ‘Trump 2028’?” I’m glad you asked because I only used this easily debunkable example to grab your attention for those cases you can’t possibly fact-check.
Imagine a deepfake of something more insidious, harder to corroborate, like the report of a terrorist attack in New York City. Or something that spreads faster, like a testimony from the CIA director that aliens exist. Or something that could turn your intimate life upside down, like fake revenge porn of yourself. Or something mundane, barely noticeable, like a spam videocall; except, suddenly, it turns into the worst nightmare possible: “Wait, that’s my daughter? Sweetie… what’s—why are you handcuffed?” “DAD, PLEASE, THEY’VE KIDNAPPED ME. PLEASE!!!”
That is terrifying. AI video (as well as other AI-generated audiovisual media) is a direct attack on a society built on trust over knowledge.
What I mean is this: we know only a bunch of things firsthand, but not many. I know who my neighbors are, where my family lives, and what my cat’s name is. Also, what color are the London plane trees down the
...This excerpt is provided for preview purposes. Full article content is available on the original publication.