Everybody Can Make DeepFakes Now [video]
youtube.com
youtube.com
Snowden and Assange both were subject to information warfare (lies is a better word) that facts don't refute. As they say "a lie is halfway around the world before the truth has its pants on."
I bet a large number of people still remember that Clinton was involved in pornography distributed from a pizza shop. Except it was not true at all. But the hatred of her lingered.
What's to happen when a normal person attempts to reveal something and then video of them surfaces showing them saying all kinds of horrible things. The scariest thing about this is you could take a picture of them from their college years and make them out to be a pedophile, and if it gets repeated enough, it will become the truth. If I admit that I had a drinking problem twenty years ago, it will "all add up." That kind of imagery is impossible for our brains to fight.
Sure some people will believe the bullshit, but I think "want to believe" is a big part of that. The sort of people who think Hillary Clinton eats babies believe it because they hate her, rather than hating her because they believe it.
I can think of 10s of stories where lives have been ruined over misrepresentation of facts in the past 5-6 years. The most prominent example of this in my mind is when a reddit ps battle faked a cheerleader having diarrhea at a game.
https://www.snopes.com/fact-check/cheerleader-accident/
She appeared on TV claiming this ruined her life. It was a single well edited photo. What can a video do? The general public still thinks faking video is outside of normal capability.
The real problem is SWGIT will take years to catch up to deepfakes. If you include case law it will likely be decades. https://en.wikipedia.org/wiki/Scientific_Working_Group_%E2%8...
Worst case: You could probably go to jail if being placed at a crime, but more likely it will just ruin your life and/or your reputation with friends and your employer.
The last 20 years the entire continent allowed PC software to be stolen and made client server, now every piece of software is spying on us because of mass computer illiteracy.
20 years ago hackers and nerds on slashdot feared Microsofts and the industries move towards hardware and software drm when the internet was just reaching the masses.
The last 20 years has been the public buying client-server drm infected software en masse.
Everything that used to be a local PC app is being pushed client-server, in such a world on privacy can exist and the public rewarded Ultima online, Everquest and world of warcraft.
For those of you who don't know, two or more PC's in a network behave as a single machine, that means whoever programs the machines in the network LITERALLY OWNS THE ENTIRE NETWORK. AKa it's the ability to rewrite the laws of every computing device on the planet and the speed of light is fast enough to take over every machine on the planet.
It's no longer merely "programming apps". It's literally stealing software en masse from the public. AKA to program now is "issuing commands", so now software can be infected with commands that take over your PC and remove your control and ownership. This is what windows 10 is about.
Some of you make complain that software was always "licensed" and I'd tell you that those laws were bribed into being by corporate lobbyists because Capitalist governments have always been ruled by corporate elites that have rubber stamped their own laws, if in doubt see the last 200 years of IP law. The public interest lost every time, that is a blow out of the public interest never winning once.
https://en.wikipedia.org/wiki/Copyright_Term_Extension_Act#/...
The fact that so many are gung-ho about "software as a service" here not grasping the huge political implications, you can't possibly audit all the software apps on google play, mobile, or steam, uplay, origin, etc. So these companies can do whatever they want and since our mind did not evolve to respond to these threats, many of the public just shrug their shoulders, because the human mind did not evolve in a technocratic society, so won't perceive nor act upon the danger.
So no, the push towards locked down filesystems you can't access in windows 10, encrypted binaries and vm's, UWP, is the big push to put bombs in honest binary blobs to make software unpreservable.
All these incentives exist because they know the public is dumb, the fortnites of the world and TF2's hat economy, told valve and big tech companies to keep pushing because the public will eat endless amounts of shit because they most of them don't care.
Imagine it's the 90's and you're buying your first PC and microsoft sells you windows, except it requires an internet connection (aka the hardware dongle) and parts of windows files and functions are held hostage on remote servers in microsofts office, most hackers of the day would not buy those PC's.
That's what "software as a service" is for many of the dolts on hackernews, its hardware dongle enabled software by way of internet and splitting all of what used to be local applications into two pieces.
It's the end of local applications on our PC's and mobile devices and most of you don't seem to be alarmed, it's the end of privacy permanently.
Because we didn't get any property rights to own our software outright with source-code, this is why steam, world of warcraft, and all new PC games like quake champions, overwatch, exist.
The end game from the early 90's was this dystopian state where all PC's on the planet are turned into dumb clients and the entire society moves to a mainframe model where no one owns their devices.
Without software ownership, our devices our useless and given the criminal way in which IP law was written for software, privacy is not coming back.
So to say the public is bright when it handed everything the nerds feared to the corporate world on a platter is bullshit.
The entire silicon valley industry is having a part at how dumb the public is, the "client to cloud" revolution is LITERALLY, stealing local apps and trapping them on company servers so they can finally turn software into property and kill the "local file" loophole on PC's to "kill piracy" by way of software as a service, the public has no idea what NTFS is.
Microsoft is now testing remote control of your file system and PC via gamepass, where companies can now set permissions to files on your PC via remote.
This is what the nerds feared in the 90's on slashdot and it has come true. The fact that hackernews is basically pro "cloud" is disturbing enough on its own.
Here's the industry celebrating the death of local user control of PC's.
https://tifca.com/wp-content/uploads/2019/09/ClienttoCloud_V...
https://twitter.com/wikileaks/status/821595404500430848
I recall at least one other tweet where they did this, but can't seem to find it right now.
In their leaked group chat logs (https://emma.best/2018/07/29/11000-messages-from-private-wik...), they behave exactly like a "para-intelligence agency" ("para" as in "paramilitary"). They happily use deceit, trickery, sockpuppetry, trolls, propaganda, and spreading of fear, doubt, and conspiracy theories to achieve their end goals, and their end goals seem to be to push things in a certain direction to achieve particular geopolitical results. Whether or not you agree with Assange's goals, he appears to have a strong "ends justify the means" mentality, much like the militaries and intelligence agencies he's opposing. Their tactics really seem identical to any intelligence agency's, at least from an information and non-kinetic perspective.
And note this was a semi-private group just to talk to their supporters. One can only imagine what their internal, top secret-classified chats and phone calls are like. I suspect those would make them look even worse than they already do, and, again, I bet they would look very similar to the most secretive of CIA meetings.
Maybe it's one of those "die or live long enough to see yourself become the villain" situations, or maybe Assange was always like this. I don't know. But they're very far from a journalistic organization, or a "Wiki", or a disinterested distributor of leaked information. They are highly political. Ironically, if they had more power or national backing [1], there would probably be a "WikiLeaks" trying to inform people about them. This sort of thing seems to be a phenomenon that repeats itself over and over again in history.
[0] https://www.theguardian.com/us-news/2017/apr/14/cia-director...
[1] Some claim they work with Russian intelligence or are even backed by Russia, but I think it's more that Russian intelligence just saw them as a useful tool for a particular operation. I don't think any nation backs them.
You call that "Pizzagate," but none of the stuff claims anything about Clinton at all as far as I can see, it's just Alefantis' weird Instagram (and former business logos) or Podesta's odd taste in artwork and word choices, which are well-documented and archived, but certainly not proof that any of them engaged in anything like child exploitation. Additionally, many of the emails can be verified via DKIM and this has been done in an HN story from the time.
I mean, sure, plenty of people have lied about Assange and WikiLeaks, obviously.
>You call that "Pizzagate," but none of the stuff claims anything about Clinton at all as far as I can see, it's just Alefantis' weird Instagram (and former business logos) or Podesta's odd taste in artwork and word choices, which are well-documented and archived, but certainly not proof that any of them engaged in anything like child exploitation.
WikiLeaks' tweet says:
>>CBS Reality Check covers Pizzagate. https://www.facebook.com/BenSwannRealityCheck/videos/1330641...
>>https://file.wikileaks.org/file/FBI-pedophile-symbols.pdf
I'm just referring to what they tweeted. Clearly this was to spread suspicion that the emails, which they leaked, might indicate Clinton is involved with pedophiles (i.e. spreading fear that Pizzagate may be real). I believe they also had one other tweet hinting at the same thing, which I remember seeing at the time, but I can't find it right now. They didn't come right out and say it explicitly, but it's an attempt at propaganda.
And if you're trying to convince me that Pizzagate is fake: yes, I absolutely know it's a bullshit unsubstantiated conspiracy theory.
It seems like your post should be directed at Assange, instead? I'm having trouble understanding what point you're trying to make.
>Additionally, many of the emails can be verified via DKIM and this has been done in an HN story from the time.
Who's disputing that? I know the emails are almost certainly legit.
> Who's disputing that? I know the emails are almost certainly legit.
Donna Brazile did so on CNN at one point, but I wasn't trying to imply that you believed that.
Your use if it is classic; applying to a broad class, specifically anyone who you percieve as not believing the same stuff you do. It's possible you know more or less about any specific subject.
It's an excellent way to shut down a conversation on the other hand, nobody wants to be labeled a "conspiracy theorist".
Here's an example, I mentioned confirmation bias hacks below, if you made a list of every use of "planar earth" on this forum, what would you find? It's not people trying to convince others that earth is a 2D surface is it?
Making unfounded accusations that certain people are guilty of some of the most heinous crimes conceivable to humanity (killing thousands of people on 9/11; raping children) isn't good. You probably wouldn't want someone to send a 60 page dossier to your employer explaining that you're a child rapist or that you faked the death of your child so you can install a totalitarian government, for example. But if you ever got on the bad side of someone with your mentality, something like that could easily happen to you.
We have a legal system and due process, and other apparatuses for when they fail. If you claim to have evidence of someone or some group committing the most horrific of crimes, truly some of the most shocking and monstrous acts in modern human history, you should be able to report such things to law enforcement, or if not possible (e.g. Abu Ghraib torture whistleblowing), inform any media outlet who can report on the accusations. Otherwise, you're libeling people.
There is a video in my profile, I made it, you can draw your own conclusions if you are interested.
Secrets, known to many, are much more common than you seem to think they are. Differences in information are similar to differences in voltage, there is more potential with a large delta.
Yes, Hillary once represented a child rapist, implied that he was guilty in an interview because she wouldn't trust lie detectors after he passed one, and got him released. But she was acting as a public defender and had no other relation to the guy. She also once helped secure the release of Laura Silsby, who had been accused of child trafficking, but there's no evidence that she had any association with her outside of that even if we simply assume that Silsby was up to no good.
Now, Bill was convicted of perjury, settled his sexual harassment suit after appealing it, and was separately accused of rape by one Juanita Broderick, but Hillary didn't have any real involvement in that, either, other than some random comments in the media calling the women involved 'bimbos' while alleging a vast right-wing conspiracy to send women to seduce Bill or stop doing laundry or something.
Oh, she also had some loose association with Anthony Wiener (AKA "Carlos Danger"), because he was married to her top aide, Huma.
Even so, the worst charge one can lay on Hillary given all this is a poor choice of words and associates, but we knew that already. I mean, she hangs out with politicians.
Nothing odd at all about the "49th most powerful person in DC" tagging an infant #hotard.
People should review the full archive, and his comments.
RE distraction: https://news.ycombinator.com/item?id=21057230
9/11 wasn't an inside job. Comet Ping Pong isn't associated with pedophilia. The guaranteed dozens of other conspiracy theories you believe (Sandy Hook, Clinton body count, FEMA death camps, probably climate change denial) are also untrue. There's a reason why belief in one conspiracy theory almost always leads to belief in most or all of them, and why others don't want to try to debate with people such as yourself.
Is it normal for a man to tag an infant picture he posted #hotard? What are #chickenlovers?
I've written many pages disputing a large number Pizzagate claims, including the ones you've brought up, on other websites in the past. It's extremely exhausting, and even more exhausting because it's so futile. No "knower of knowledge they don't want you to know" ever thinks or accepts they might actually be wrong about any of their extraordinary claims.
If someone could compensate me for my time, I'd definitely be willing to spend many days disputing every single criminal plot you wish to present. I think it's very likely there are at least a dozen. Again, if you believe in one of them, you likely believe in many others - that's confirmation bias. No, I shouldn't have said "guaranteed to believe", but I think it's highly probable.
Is it normal for a man to tag an infant picture he posted #hotard? What are #chickenlovers?
DeepFakes just seem to be adding video to the list of fakeable things, which already included photos and audio.
As a side note, I am less concerned with the use of deep fakes to frame (civil or criminal) a person, it is more the potential for an attorney to put big wholes in a jury’s trust of video evidence by showing a faked image as counterproof and then flat out telling jurors that one is fake and therefore the other side can’t prove it’s case.
Caveat -> this doesn’t apply to police generated evidence, because a majority of jurors, to a statistical certainty barring exceptional circumstances, ALWAYS believe cops.
A simple but possibly useful scenario could be at the time of recording a video, immediately storing a hash of its content on a blockchain somewhere to validate its authenticity later (if needed). Imagine if iPhone or Android did this by default whenever a video or picture were taken.
Deepfakes are just another prong on the same trident. I don't see how they are any worse than the screenshots of bullshit headlines that already go viral on Reddit/Twitter.
Hell, just think of all the suffering that's been caused by people believing in some random stuff written by other humans in an old religious book without questioning it. And people think it's deepfakes that might be civ-ending?
But that suffering was civ-ending for some civilizations. I don't really care if some civilizations thereafter survive and go on to forget about this impending crisis. I care that MY civilization survives.
I also don't care to survive in a civilization if it is ruled by suffering.
It will boil back down to basic philosophy, questioning everything. Our senses have always lied to us, just as our emotions and motivations. Maybe we'll just learn to be better at discerning the lies.
People do understand that fake videos can happen, but they also believe they require a great deal of effort. That means they might think a video of Trump is fake because his enemies have the resources to fake it, but they won't think the same way about a video of a member of the public or someone they know.
I'm quite sure that if 99% of people outside of the tech industry were shown a deep fake video of a co-worker in a porn clip they would believe it was real.
Because we will finally stop trusting videos and images as a reliable proof, when they clearly aren't all the time.
People already know videos and images can be faked, but they still trust whatever they want to be true as a post-hoc validation of their existing prejudice, not as the result of a rational attempt to judge evidence on its own merits.
Just look on Youtube for various "paranormal" videos, things like "real proof" of reptilians, ghosts, aliens, etc. Most of those videos are laughably fake looking, but people who already believe in these things take them as gospel, anyway.
Politically speaking, look at Pizzagate as another example. Whether or not you believe the "evidence" tends to depend on what you already believe about Democrats/progressives/leftists/the Clintons.
is it even illegal? even if it is, could we do anything if they operated out of Russia or the middle East?
[0]https://www.forbes.com/sites/steveolenski/2017/06/12/the-bra...
[1]https://www.wired.com/story/messy-legal-fight-to-bring-celeb...
https://m.youtube.com/channel/UCKpH0CKltc73e4wh0_pgL3g/video...
Here’s a pretty well done one in which Bill Hader does impressions, during which his face is replaced with the one of the people he’s doing an impression of: https://m.youtube.com/watch?v=kjI-JaRWG7s
Saw something like this in a Neal Stephenson book I believe.
It’s like AlphaZero playing against itself and you hoping that the “defensive” AlphaZero is able to win the game every time.
In reality, each side will win sometimes but will have absolutely no way of explaining why or how. So we will have no idea anymore what is fake or not, just whether the detector said it was for some reason that time.
Basically, it’s game over very soon after you do this arms race. See CAPTCHAs being solved etc.
Mission accomplished? https://xkcd.com/810/
It would also allow any device manufacturer to sign anything they like, as well as anyone who can coerce a device manufacturer to do so.
Thanks for replying though, this does help me understand the challenges in a system like this.
For example, it was discovered that it's possible to extract keys from Intel SGX enclaves using certain speculative execution vulnerabilities. Intel SGX predates Spectre. It isn't a category of vulnerability they knew existed when they were designing it.
Vulnerabilities are regularly discovered in almost everything, iPhones included. Diligent vendors are quick to patch them, but an attacker only needs to wait until the next vulnerability is discovered and then extract the signing keys from a device that hasn't been patched yet.
You also have no way of knowing which keys they are -- if a million devices of a particular model have a known vulnerability then any attacker could extract the keys from any of them, and even blacklisting all of them (which would tend to dissatisfy their innocent owners) still wouldn't save you from an attacker using an unpublished vulnerability against a device you don't even know is vulnerable.
To put this another way, this is basically the same class of technology that Hollywood uses for DRM. Now, how many Hollywood movies can you say have not been pirated by anyone?
SOTA model training now costs up to 7 figures. Creating deep fakes that couldn't be detected by some 8 figure SOTA model might not be worth it; definitely no regular Joe Deep Learning researcher could make one at some point. Then it will be done only for high value targets by some large corporate or government bodies.
> AWS, Facebook, Microsoft, the Partnership on AI’s Media Integrity Steering Committee, and academics have come together to build the Deepfake Detection Challenge (DFDC). The goal of the challenge is to spur researchers around the world to build innovative new technologies that can help detect deepfakes and manipulated media.
This has a prize pool of $1 million.
The tools for digital signing already exist and are used in niche industries but not really something everyday people are concerned about. It's more a social awareness problem and this will come with better tools, integrated with the apps and channels ordinary people use daily so that it is something that happens in the background.
In addition to digital signing, public distributed ledgers can establish provenance or a chain of ownership for signed content, so digital content that is altered and redistributed can be sourced back to its origin more accurately.
Ordinary people with the right tools built into their devices will be able to see very easily "Is it in the chain?" based on a simple enquiry to the ledger, and if the answer is no, then it's untrustworthy.
If the answer is yes, then they will ask "Does trace to the origin?" and if it does not, or the origin isn't signed by the alleged owner, then it is untrustworthy.
That's one possible way crypto can help defeat disinformation / fake news content. The underlying techniques exist but there is a lot of work to do to bind it to everyday use.
Not that I disagree, but the first thing that came to mind when I read this part of your comment was that this approach reinforces the Monopoly of governments and media companies on the flow of information. On the one hand it's great that we could protect in this way from unverified, "not-in-the-chain" messages, but on the other hand being a contrarian, producing and coming into contact with information outside of the chain, not necessarily really fake videos, but other types of information is crucial to social evolution and civil discourse.
The "public" and "distributed" would be major priorities of mine in the scenario you describe.
Their photo of the photo now has a valid cryptographic signature that proves it's real and trustworthy.