How Do We Stop Deepfakes from Tricking Juries? | Hacker News Reader

How Do We Stop Deepfakes from Tricking Juries?

gizmodo.com

5 pointsmostcallmeyt5 comments

Any reasonable defence lawyer would simply submit "evidence" showing that the judge or prosecutor actually committed the crime. The point being to make it obvious that photos and footage can no longer be trusted.

The p(doom) by deepfakes against epistemic reality is a nightmare fuel. This is one of the very few things that should be uninvented and chucked into either the abyss of an ocean or into a volcano.

If you think that's bad, wait until you hear about Photoshop. You can actually control the presentation of the image. This dispicable company also supplies criminals with the popular industry-grade deepfake software "After Effects" that has been used to composite fabricated evidence against real life footage.

I think the threat deepfakes pose to the current legal system has been solved since computers became a household item. You're being hyperbolic.

Comparing the utter scale of the problem this time around to things such as photoshop shows just how little you understand the sheer magnitude of this problem. That's even worse than simply comparing apples to oranges.

Photoshop and similar applications did allow misinformation to be made. But not even 0.01% as much as AI-powered deepfakes as we see today. Photoshop took time, skill and had limitations. The worst someone could really do before was photoshop your head onto the body of someone else.

Now videos that look 100% real can be generated at lightning speed with no effort. Think of a celebrity, or a politician. Half a decade ago if you saw a photo of them you could be pretty certain it actually was them. Now you can’t. Half a decade if you heard an audio clip of a particular person confessing to something you could be almost certain it was them. Now you can’t. Videos and audio used to be pretty reliable before. Now they’re useless. AI deepfakes makes misinformation 1000X worse then Photoshop ever did.

It's sensible to think that it is possibly one of the reasons of the Great Filter.

If anything, it would seem to make convicting actual criminals much harder since the ubiquitous use of AI renders all photo and video evidence useless to prove a crime “beyond a reasonable doubt.”

Previously, of course, those things could be faked but because of the skill required to do so, it would make it less plausible to question their legitimacy “beyond the reasonable doubt.” There’s also the fact of that would have required a human to do so.

Now, images can be modified at their source programmatically which throws questionable chain of custody into the mix as well.

Reply on news.ycombinator.com