Deepfakes: MIT brings Nixon's Apollo disaster speech to life • Hacker News Board

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#141

Post by gonehome » Fri, Nov 22, 2019, 9:25 PM UTC

pc86 wrote: post #131

canada_dry wrote: post #58

Deep fakes are yet another prime example of how the writing is on the wall (i.e. it's clear that this technology can/will be used for nefarious purposes - esp. by our enemies) and yet mitigations - through regulation and enhanced standards - won't be pursued until long after the horses have left the barn. I don't know how/what form this should take, but an older analogy might be how colour photocopier manufacturers i…

This comment is so egregiously out of touch I hesitate to even respond at all, let alone to do so civilly, however in the interest of discussion I'm going to try to. Which "enemy" do you envision would use deep fakes for a nefarious purpose, yet will stop short once they realize there are - gasp! - regulations around them? And as far as I'm aware, at least in the United States, there are no regulations requiring the…

You said you'd try to respond civilly, but your comment is openly hostile and this takes away from your valid core point.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#142

Post by Nasrudith » Fri, Nov 22, 2019, 9:29 PM UTC

surfsvammel wrote: post #79

We need ways to sign or put signatures in videos that proves their authenticity. Not sure how such a thing would work though... I guess, in the world of blockchain, we could guarantee it’s origin at least.

The common mistake with signing schemes is assuming that it actually means anything on its own. There is no need to it.

That would only make sense if you want to mark something as offical and have authorative sourcing. It would have a place for chains of command and actually signed off on orders but not evidence that the content is true.

An offically signed video of the president punching Adolf Hitler in the groin, High Fiving George Washington, and then flying off into space could be cryptographically valid although obviously fake. It would only say that the President actually approved this but not that it was real.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#143

Post by driverdan » Fri, Nov 22, 2019, 9:38 PM UTC

TimMurnaghan wrote: post #2

It still looks wrong. There's still some of the floating/stuck-on head feeling about it. Or is that just because we already know it's fake from the headline? (... and the way that the flag moves is totally wrong .../s).

The wider shot definitely had an uncanny valley feeling for me. His head movement seemed unnatural.

The tight shot, however, was really good. I'm not sure I'd think it was fake if I wasn't primed for it.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#144

Post by kragen » Fri, Nov 22, 2019, 9:42 PM UTC

> "Deepfakes can be used for many of the things we already know," [co-director of the Nixon film Francesca] Panetta says, "but also to create kind of alternative histories or have the potential to kind of rewrite history as well."

Movies, from Hollywood and elsewhere, have been convincing people of rewritten histories since their inception. How well known is the story of the real Spartacus compared to the Hollywood Spartacus? How many people's vision of the antebellum US South came from Gone With The Wind rather than from historical documents and interviews with former slaves and slave-owners? Birth of a Nation inspired the revival of a terrorist organization that lasted decades; Triumph of the Will painted the Nazis' rise as a matter of noble heroism triumphing over cowardice. Last week I argued with some epistemologically incompetent person who wanted me to watch an anti-vaccine movie about Gardasil, apparently unaware that YouTube videos are not really a publication venue used by medical researchers.

So, what should we do about it? Well, the Soviets had an answer: since movies were so powerful, people would be carefully vetted before they got access to the equipment needed to make them, and if someone made a movie with harmful contents anyway, they would go to GULAG. Is that the solution we want?

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#145

Post by TearsInTheRain » Fri, Nov 22, 2019, 9:42 PM UTC

ISL wrote: post #48

3fe9a03ccd14ca5 wrote: post #22

I find it amusing that Congress has suddenly become so interested in the dangers of deep fakes. When they started going on and on about the dangers I thought “why is this dude so worried?”. Then of course the Epstein stuff becomes news[1] months later and the cynical part of me can’t help but think they’re looking for a defense if they ever partied at his island or ranch. 1. https://nypost.com/2019/11/18/jeffrey-epst…

Deepfakes are readily understandable as a problem for politicians. Politicians are experts in public messaging; this technology has the ability to alter previously-immutable messages.

Is there any benefit to deepfake technology besides maybe entertainment?

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#146

Post by cwkoss » Fri, Nov 22, 2019, 9:45 PM UTC

michaelt wrote: post #122

Reason077 wrote: post #113

Earlier quoted context omitted.

Could blockchain give us a secure audit trail for digital video clips?

The blockchain could provide a timestamp, but it's hard to see how that would help. Every type of video authentication system I've ever heard proposed can be defeated by simply pointing a 'trusted' video camera at a screen playing a fake video.

I was chatting with someone about making deep-fake-resistant video. An interesting solution that came up was making artifacts that would be hard to simulate: ex. surround the speaker is a swirling vortex of confetti, where words have a minimal but detectable effects on the air currents.

Video could be 'authenticated' by post facto tracking of particles to validate both continuity and plausible velocity deltas based on generated model of air currents. It wouldn't eliminate the possibility of deepfakes, but would drastically increase cost of production.

Similarly, I predict that cameras will increase in resolution past the point of utility for human vision: tetrapixel camera video would be extremely expensive to produce without leaving tiny artifacts.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#147

Post by Bedon292 » Fri, Nov 22, 2019, 9:46 PM UTC

3fe9a03ccd14ca5 wrote: post #63

ouid wrote: post #52

Earlier quoted context omitted.

Of all the places where you could find "epstein news" you choose to link the New York Post.

I would honestly love to link to a trusted MSM report, but these outlets have been burying Epstein stories[1]. It's awful. 1. https://www.newsweek.com/abc-jeffrey-epstein-story-amy-robac...

Except it "didn't air because we could not obtain sufficient corroborating evidence to meet ABC's editorial standards about her allegations." This should be an article about journalistic integrity, and ensuring we don't accuse people of things without corroborating evidence. Just because more evidence came out after that point, does not mean it was wrong to require more evidence at the time. I cannot see how this is anything other than a good thing.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#148

Post by schainks » Fri, Nov 22, 2019, 9:47 PM UTC

Sounds like a good time to create a honeypot deepfakes service. Could charge a handsome fee for it, too, then authenticate deepfakes funded by political parties.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#149

Post by pc86 » Fri, Nov 22, 2019, 9:57 PM UTC

Bartweiss wrote: post #107

jdietrich wrote: post #90

Earlier quoted context omitted.

> Deepfakes is just a scary new word for what people would call "CGI" or "VFX". "Ryzen 3950x" is just a scary new word for "z80" or "6502". Deepfakes don't provide any fundamentally new capability, but they do reduce the cost and time required to create a convincing fake video by multiple orders of magnitude. That completely changes the threat model, from "our propaganda rival occasionally releases a fake video that…

> Deepfakes don't provide any fundamentally new capability I'm not sure this is entirely true, although your larger point is quite good. People have been faking video since the dawn of video, yes. But for several decades, a clear, high-res video of a specific person has been pretty much inviolate. Faking a steady closeup of a president or famous actress would have been unthinkable - no VFX hoax ever convincingly impe…

The point is that nobody talks about making altered photographs illegal, or having federal regulations around altering photographs, or anything remotely as ridiculous. But because it's video everybody is losing their minds.

Re: Deepfakes: MIT brings Nixon's Apollo disaster speech to life

#150

Post by cwkoss » Fri, Nov 22, 2019, 10:01 PM UTC

basseq wrote: post #96

OK, forget the implications of low-cost and convincing fake video footage for nefarious purposes. "Fake videos"—e.g., CGI—have been around forever! I don't suddenly think Star Wars is real because I saw a video of it. At the end of the day, what cool tech for legitimate purposes! Hollywood VFX, training video or PR customization, etc. Nixon giving this speech is a cool look into "what-if", without the moral burden of…

The point is that society currently takes video at face value (pardon the pun)

A clever hacker could replace the face on security footage of someone committing a crime with yours, put that file back on to security cam dvr, removing metadata that indicates the file was ever modified. Let the police retrieve the footage from the dvr. Do you think a good lawyer could get you off the charges for committing that crime, if you didn't have an alibi?

Evidence standards around video will need to change soon, and society isn't ready for this shift currently.