The White House wants to ‘cryptographically verify’ videos of Joe Biden so viewers don’t mistake them for AI deepfakes::Biden’s AI advisor Ben Buchanan said a method of clearly verifying White House releases is “in the works.”
The White House wants to ‘cryptographically verify’ videos of Joe Biden so viewers don’t mistake them for AI deepfakes::Biden’s AI advisor Ben Buchanan said a method of clearly verifying White House releases is “in the works.”
Would someone have a high level overview or ELI5 of what this would look like, especially for the average user. Would we need special apps to verify it? How would it work for stuff posted to social media
linking an article is also ok :)
Depending on the implementation, there are two cryptographic functions that might be used (perhaps in conjunction):
Cryptographic hash: An arbitrary amount of data (like a video file) is used to create a “hash”—a shorter, (effectively) unique text string. Anyone can run the file through the same function to see if it produces the same hash; if even a single bit of the file is changed, the hash will be completely different and you’ll know the data was altered.
Public key cryptography: A pair of keys are created, one of which can only encrypt data (but can’t decrypt its own output), and the other, “public” key can only decrypt data that was encrypted by the first key. Users (like the White House) can post their public key on their website; then if a subsequent message purporting to come from that user can be decrypted using their public key, it proves it came from them.
A note on this. There are other videos that will hash to the same value as a legitimate video. Finding one that is coherent is extraordinarily difficult. Maybe a state actor could do it?
But for practical purposes, it’ll do the job. Hell, if a doctored video with the same hash comes out, the White House could just say no, we punished this one, and that alone would be remarkable.
You’d need to find one that was not just coherent, but that looked convincing and differed in a way that was useful to you—and that likely wouldn’t be guaranteed, even theoretically.
Even for a 4096 bit hash (which isn’t used afaik, usually only 1024 bit is used (but this could be outdated)), you only need to change 4096 bits on average. Even for a still 1080p image, that’s 1920x1080 pixels. If you change the least significant bit of each color channel, you get 6,220,800 bits you can change within anyone noticing. That means on average there are 1,518 identical-looking variations of any image with a given 4096 bit hash, on average. This goes down a lot when you factor in compression: those least significant bits aren’t going to stay the same. But using a video brings it up by orders of magnitude: rather than one image, you can tweak colors in every frame The difficulty doesn’t come from the existence, it comes because you need to check 2⁵¹² = 10¹⁵⁴ different images to guarantee you’ll find a match. Hash functions are designed to take a while to compute, so you’d have to run a supercomputer for an extremely long time to brute force a hash collision
This concept is known as ‘collision’ in cryptography. While technically true for weaker key sizes, there are entire fields of mathematics dedicated to probably ensuring collisions are cosmically unlikely. MD5 and SHA-1 have a small enough key space for collisions to be intentionally generated in a reasonable timeframe, which is why they have been deprecated for several years.
To my knowledge, SHA-2 with sufficiently large key size (2048) is still okay within the scope of modern computing, but beyond that, you’ll want to use Dilithium or Kyber CRYSTALS for quantum resistance.
The best way this could be handled is a green check mark near the video that you could click on it and it would give you all the meta data of the video (location, time, source, etc) with a digital signature (what would look like a random string of text) that you could click on and your browser would show you the chain of trust, where the signature came from, that it’s valid, probably the manufacturer of the equipment it was recorded on, etc.
Just make sure the check mark is outside the video.
The issue is making that green check mark hard to fake for bad actors. Https works because it is verified by the browser itself, outside the display area of the page. Unless all sites begin relying on a media player packed into the browser itself, if the verification even appears to be part of the webpage, it could be faked.
Hope verification gets built in to operating systems as compromised applications present a risk too.
But I’m sure a crook would build a MAGA Verifier since you can’t trust liberal Apple/Microsoft technology.
The only thing that comes to mind is something that forces interactivity outside the browser display area; out of the reach of Javascript and CSS. Something that would work for both mobile and desktop would be a toolbar icon that is a target for drag-and-drop. Drag the movie or image to the “verify this” target, and you get a dialogue or notification outside the display area. As a bonus, it can double for verifying TLS on hyperlinks while we’re at it.
Edit: a toolbar icon that’s draggable to the image/movie/link should also work the same. Probably easier for mobile users too.
Adobe is actually one of the leading actors in this field, take a look at the Content Authenticity Initiative (https://contentauthenticity.org/)
Like the other person said, it’s based on cryptographic hashing and signing. Basically the standard would embed metadata into the image.
For the average end-user, it would look like “https”. You would not have to know anything about the technical background. Your browser or other media player would display a little icon showing that the media is verified by some trusted institution and you could learn more with a click.
In practice, I see some challenges. You could already go to the source via https, EG whitehouse.gov, and verify it that way. An additional benefit exists only if you can verify media that have been re-uploaded elsewhere. Now the user needs to check that the media was not just signed by someone (EG whitehouse.gov. ru), but if it was really signed by the right institution.
As someone points out above, this just gives them the power to not authenticate real videos that make them look bad…
I honestly feel strategies like this should be mitigated by technically savvy journalism, or even citizen journalism. 3rd parties can sign and redistribute media in the public domain, vouching for their origin. While that doesn’t cover all the unsigned copies in existence, it provides a foothold for more sophisticated verification mechanisms like a “tineye” style search for media origin.
Videos by third parties, like Trump’s pussy grabber clip, would obviously have to be signed by them. After having thought about it, I believe this is a non-starter.
It just won’t be as good as https. Such a signing scheme only makes sense if the media is shared away from the original website. That means you can’t just take a quick look at the address bar to make sure you are not getting phished. That doesn’t work if it could be any news agency. You have to make sure that the signer is really a trusted agency and not some scammy lookalike. That takes too much care for casual use, which defeats the purpose.
Also, news agencies don’t have much of an incentive to allow sharing their media. Any cryptographic signature would only make sense for them if directs users to their site, where they can make money. Maybe the potential for more clicks - basically a kind of clickable watermark on media - could make this take off.
It needs some kind of handler, but we mostly have those in place. A web browser could be the handler for instance. A web browser has the green dot on the upper left, telling you a page is secure, that https is on and valid. This could work like that, the browser can verify the video and display a green or red dot in the corner, the user could just mouse over it/tap on it to see who it’s verified to be from. But it’s up to the user to mouse over it and check if it says whitehouse.gov or dr-evil-mwahahaha.biz
TL;DR: one day the user will see an overlay or notification that shows an image/movie is verified as from a known source. No extra software required.
Honestly, I can see this working great in future web browsers. Much like the padlock in the URL bar, we could see something on images that are verified. The image could display a padlock in the lower-left corner or something, along with the name of the source, demonstrating that it’s a securely verified asset. “Normal” images would be unaffected. The big problem is how to put something on the page that cannot be faked by other means.
It’s a little more complicated for software like phone apps for X or Facebook, but doable. The problem is that those products must choose to add this feature. Hopefully, losing reputation to being swamped with unverifiable media will be motivation enough to do so.
The underlying verification process is complex, but should be similar to existing technology (e.g. GPG). The key is that images and movies typically contain a “scratch pad” area in the file for miscellaneous stuff (metadata). This is where the image’s author can add a cryptographic signature for the file itself. The user would never even know it’s there.
Probably you’d notice a bit of extra time posting for the signature to be added, but that’s about it, the responsibility for verifying the signature would fall to the owners of the social media site and in the circumstances where someone asks for a verification, basically imagine it as a libel case on fast forward, you file a claim saying “I never said that”, they check signatures, they shrug and press the delete button and erase the post, crossposts, and if it’s really good screencap posts and those crossposts of the thing you did not say but is still being attributed falsely to your account or person.
It basically gives absolute control of a person’s own image and voice to themself, unless a piece of media is provable to have been made with that person’s consent, or by that person themself, it can be wiped from the internet no trouble.
Where it comes to second party posters, news agencies and such, it’d be more complicated but more or less the same, with the added step that a news agency may be required to provide some supporting evidence that what they said is not some kind of misrepresentation or such as the offended party filing the takedown might be trying to insist for the sake of their public image.
Of course there could still be a YouTube “Stats for Nerds”-esque addin to the options tab on a given post that allows you to sign-check it against the account it’s attributing something to, and a verified account system could be developed that adds a layer of signing that specifically identifies a published account, like say for prominent news reporters/politicians/cultural leaders/celebrities, that get into their own feed so you can look at them or not depending on how ya be feelin’ that particular scroll session.