Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Investigating AI Manipulation in Viral Chinese Paraglider Video

blog.hyperknot.com

Post reply on HN
Print view View original

51–60 of 62 posts

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#51

Post by gosub100 »

shigawire wrote: post #32
webdoodle wrote: post #18

AI has finally reached a point at which it can fake videos well enough that most people will believe it. Just in time for the fake Epstein suicide video! Epstein didn't kill himself, nor is he dead. He's living out his retirement on another pedo island, as a reward for a kompromate job well done.

Based on what? I get the conspiracy that he was killed by people with something to cover up. But what is the motivation for rewarding him? Unless you suppose he has a very compelling dead man's switch?

Based on the fact that the security cameras "weren't working" and the guards were "sleeping" at the moment he did it. When one of the highest profile criminals was in their unit. Bull crap.

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#52

Post by qingcharles »

jonplackett wrote: post #9

Whatever the outcome of this specific case, we are clearly at a tipping point in history - it’s now basically impossible to say for certain if a fake video is real, and potentially even more problematic - also impossible to say for sure that a real video is not fake. The whole concept of ‘video evidence’ is going up in smoke.

Most of us on here have an acutely-tuned spidey-sense when it comes to detecting AI videos, but several times recently I've found myself completely unable to tell if a video is real or fake. The issue is: if it's real, then you can spend a ton of time trying to determine it's fake, when it's not, and still not be sure.

Either way, we're cooked.

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#53

Post by wongarsu »

amlib wrote: post #23
kjkjadksj wrote: post #20

Earlier quoted context omitted.

Uhh, CGI is decades old now. Videos without sufficient provenance have always been suspect. Even beforehand you could do plenty of tomfoolery in the cutting room.

But that required lots of expertise and time to fabricate anything remotely convincing. We may now be close to an era where fakery at that level can be fabricated at industrial scale by any individuals and, most importantly, by powerful entities.

Powerful entities have always been able to hire VFX artists. Preferably fresh out of film school for 20% above industry wage and a sense of purpose. For the rest of us there's Fiverr. Some guy in Indonesia is unlikely to investigate whether he's editing part of a prank video or fabricating evidence.

Fabricating video absolutely became easier. But it's not so much about your video skills, you can buy those. It's about your organization and planning skills. Even a hilariously incompetent and unorganized entity with a lack of foresight can now fabricate reasonably convincing video

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#54

Post by gosub100 »

nullc wrote: post #35
mattlondon wrote: post #34

Earlier quoted context omitted.

Not for a second that I believe the conspiracy, but he would be rewarded for taking the blame post-mortem so a lot of other rich people don't .

Epstein was just a con man, he wove a fog of mystery around himself as part of the con and people are still being duped by it to this day.

Why was he photographed on his island wearing a cap that said "missed" ?

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#55

Post by jq-r »

qingcharles wrote: post #52
jonplackett wrote: post #9

Whatever the outcome of this specific case, we are clearly at a tipping point in history - it’s now basically impossible to say for certain if a fake video is real, and potentially even more problematic - also impossible to say for sure that a real video is not fake. The whole concept of ‘video evidence’ is going up in smoke.

Most of us on here have an acutely-tuned spidey-sense when it comes to detecting AI videos, but several times recently I've found myself completely unable to tell if a video is real or fake. The issue is: if it's real, then you can spend a ton of time trying to determine it's fake, when it's not, and still not be sure. Either way, we're cooked.

Ars Technica just did a nice article on purely generated videos and some of them look completely real to my eyes.

https://arstechnica.com/ai/2025/05/ai-video-just-took-a-star...

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#57

Post by dingnuts »

Molitor5901 wrote: post #6
ribfeast wrote: post #4

Earlier quoted context omitted.

Can be to validate the technology before using it for something more nefarious.

I think it's validation. Consider if you were wanting to perfect AI to be very hard to distinguish, what better way than to crowd source it by putting on social media? The fear is when it's perfect and someone puts you in the middle of a very illegal situation; prove it wasn't you.

> The fear is when it's perfect and someone puts you in the middle of a very illegal situation; prove it wasn't you.

At least in the US, while the law is followed, this is not legal, the State will have to prove it wasn't AI. They're required to assume it's not you, and prove that it is.

The system is flawed of course and I'm not saying it always plays out like this in the real world, but the more we remember and repeat the laws they are supposed to follow, I think anyway, the greater chance we can hold those in power to the laws to which they are supposed to be held

Re: Investigating AI Manipulation in Viral Chinese Paraglider Video

#59

Post by butlike »

ribfeast wrote: post #4
tartoran wrote: post #2

I guess we'll have to take everything with a grain of salt from now on as any evidence can be faked with AI. I'm likely naive since I don't understand what the point of faking this was. Clout chasing? Monetizing this somehow?

Can be to validate the technology before using it for something more nefarious.

Yup. Defining the overton window for emerging technologies
Post reply on HN