Hacker News Reader

Comment by photochemsyn | Hacker News Reader

What are some ways to verify video integrity?

1) Multiple camera angles. A simultaneous left / front / right view of a person speaking would be much harder to fake.

2) Higher-resolution video and high-quality audio are harder to fake, although this might not be a problem for entities with sufficient resources and motivation.

3) Attribution to a real person. While this raises problems for people who want to remain anonymous due to very reasonable fears of retaliation, it also makes it harder to push fakes out to social media.

Note many fake 'breaking news' videos are just stiched together old footage, no AI deepfake required. This has been noticeable across multiple platforms wrt Ukraine, for example:

https://www.bbc.com/news/60867414

View on HN

Unfortunately, while those are good ideas, they’re also cat and mouse games. It’s a bit like saying that client side anti-cheat software makes it harder to cheat in videogames. It does, but you’re looking for an unbeatable test, which doesn’t exist.

Nothing exemplified this problem more than the recent Twitter takeover. On tiktok, fake news was running rampant, and I saw it happen in real time. At one point someone said that a Twitter employee banned Elon on his way out, while showing a screenshot of the “eIon musk” fake account: https://www.tiktok.com/t/ZTRx6G9mG/

There were also actual deepfakes of Elon’s voice that fooled me. I wish I saved it, since it’s highly relevant to this thread. But it only became clear it was a fake when it started saying ridiculous things. They even had a video of Elon talking, and they didn’t have to bother syncing his lips with the audio. Viral clips just need a veneer of authenticity to short circuit our monkey brains.

It wasn’t till I fell for that deepfake that I started realizing how much of a problem we might be facing. The next few presidential candidates might need to put out statements saying “X wasn’t true,” while the majority of people already believe X because they saw it happen live on tiktok.

Yeah, that's all true. This might lead to a revival of professional journalism and actual reporting, however - i.e. if a known reporter with a good track record writes, 'today this presidential candidate said X, and I heard it myself', then that might become a more reliable information source than any audio or video clip.

Ah, but what happens when a deepfake video of that professional journalist starts going round saying 'today this presidential candidate said X, and I heard it myself' that matches the previous deepfake?

I guess the counter (which I think you're pointing out) to that is to publish text to a known-trusted website rather than a video, but I wonder how many folks who get their "news" from TikTok would read it.

That’s when real verified accounts gain value but since now anybody can be verified, it is going to become a problem.

Uhm, I am not sure I get the relationship: if anyone can get verified, that's only a win.

By "verified", I assume Twitter has verified that their identity on Twitter matches their real life identity, and I don't see a problem with anyone having an ability to claim that. What's wrong with Jane Random being verified that she _is_ Jane Random?

If people are basing their trust not on someone's identity and history, but on their celebrity (blue mark so far), that's where the problem is.

The problem arises where an account gets verified as one thing and then changes to become another, this happens all the time with trolls who can make/buy/extort verified accounts.

They then spoof being the real person, and verification then means absolutely nothing cause they are functionally identical to a bypasser.

By having verification be really hard to get, trolls have a lot harder time getting accounts and the problem is smaller in scope.

Allowing identity changes without requiring reverification to keep the "verified" status sounds like a problem with an obvious solution: Require revalidation, or remove the mark.

Having initial validation accessible is not the issue here.

Exactly: I am totally perplexed at all the "see what verifying anyone will result in", instead of "your verification is broken" claims.

It's like allowing verified emails to be changed to another address without reverifying.

I mean, doh.

All the evidence points towwards the opposite: it's more profitable to simply provide entertaining lies. Does nobody remember the Alex Jones fiasco?

It's very much a stretch to compare Alex Jones to normal journalists, as much as most people seem to dislike journalists on this site.

Ah, then I was got by that fake twitter ban joke and didn't realize it till now. In my defense, it's a cute little meaningless stunt that I'd like to imagine myself doing in similar circumstances, though upon a bit of thought it seems unlikely since it could cost you a severance that might be pretty valuable - there a plenty of strings attached to those to keep the severees in line.

There's been a phenomenon of verified accounts basically doing ""suicide by cop"": changing their display name and profile pic to match Elon Musk and then doing a bunch of joke tweets as him, which are great fun until they are caught and banned.

I've seen Youtube shorts featuring deepfakes of Elon promoting some cryptocurrency scam a few weeks ago.

Video integrity is an idea we have to let go of. Audio is robustly fakeable at this point, video is coming up to the threshold fast.

What we can have is attestation and reputation, then provenance.

It starts with a container format that allows signatures of both the entire video with audio, audio only, and video only, along with reasonable segments (three seconds?) of each of these.

Excerpts of any of these can be transcluded into another such container. The editor can transform it in any way, the resulting display doesn't have to match the particulars of the excerpt (captioning, noise filtering, color correction) but the excerpt is itself included, or at least the signature.

It would be useful for some people to have cameras which do this at the moment of recording, along with GPS, cell location data, and so on, to get really bulletproof documentary footage.

I don't want that built into my phone, necessarily. As a container format and ffmpeg style tool, this would support anonymous attestation which can be demonstrated by the keyholder at any future time. As a built-in subpoena-able feature of my phone or mirrorless camera? Hmm, probably don't want that. If it snitches proactively, not a chance in hell.

Even just a verifiable time stamp would be a big win.

One way to get a verifiable time stamp would be to show knowledge of an unpredictable but well-known thing in the past (providing a lower bound), then to create the same thing yourself (providing an upper bound). I wrote about it here: https://news.ycombinator.com/item?id=33558566.

Your comment in particular made me feel like it was worth posting!

Building attestation into hardware sounds like a good idea in theory, but doesn't actually work. All it tells you is "this is what the camera saw", it doesn't actually tell you that what the camera saw was real. You can just point your camera at a high resolution screen playing a doctored video, and so sign whatever doctored content you like. Or take a video of a real faked scene with actors.

Of course it works, it works trivially.

We mustn't confuse attestation with reputation, nor even the combination of attestation and reputation with truth.

Attestation is nothing more than someone staking their reputation on the veracity of what they're sharing with the world.

If I'm being frank, I thought this was too obvious to need stressing. If the camera is pointed at David Copperfield, cryptography can't make the lens show you reality.

Attestation solves a lot of problems, but like my first sentence said, video integrity isn't a problem which can be solved anymore, if it ever was.

The trusted GPS problem might be solved by actually turning GPS upside down and having the device emit the signals instead of receiving them—it’s known as reverse or inverted GPS.

I’ve written a blog post that addresses many of problems, and your comment was one of the reasons I posted it! You can find it at https://news.ycombinator.com/item?id=33558566

The founder of Bellingcat discusses this in his recent book.

He mentions things like "where/when did this happen", where were the people/things in the video at that time, which source did this arise from. Basically there's a whole web of context to any event, which is why the breaking news is the best place for this, but the long term viability is much harder.

He also mentions some NYT tech for timestamping their videos as they record them and to detect manipulation.

Can you say more about this tech? I'm curious now :)

https://www.newsprovenanceproject.com/

Mentioned on page 206, "The Perils and Opportunities of Artificial Intelligence" of the book.

https://c2pa.org/ seems to be the latest iteration of it.

> What is the Coalition for Content Provenance and Authenticity (C2PA)?

> The C2PA is a Joint Development Foundation project to collectively build an end-to-end open technical standard to provide publishers, creators, and consumers with opt-in, flexible ways to understand the authenticity and provenance of different types of media. C2PA opposes efforts to make content provenance measures mandatory.

It would require hardware changes, but we already have all the required cryptographic primitives to mostly mitigate these concerns.

Trusted timestamping can attest that a given piece of media was not created after a given instance in time.

Trusted hardware running trusted code, combined with remote attestation, can offer a high-degree of assurance that a given video clip is not a product of CGI/ML.

Digital signatures by the entity responsible for a video clip's capture/creation mitigates the creation of media not sanctioned by that entity. Subjects in a particular piece of media could additionally apply their digital signatures to further increase trust in its authenticity. (This is effectively a cryptographically enforced version of your third point.)

This isn't a silver bullet, and suffers all the pitfalls normally encountered in the implementation of cryptographic schemes.

I can just project a video into a camera's lens...

Fun fact: the original Apollo TV broadcast from the surface of the moon was "slow scan", with a different resolution and framerate to broadcast TV. The conversion process was to display it on a suitable slow-scan TV (long retention phosphor) and .. point a terrestrial camera at the TV.

Sure, and if you eventually get caught, nobody would trust anything that you've digitally signed.

You can say that just as well about a world without the camera hocus pocus part.

In practice, not really.

The key difference is that digital signatures allow revocation-of-trust to be an effortless, automatic and delegatable process. As opposed to the current process that involves individual (and intrinsically limited) knowledge of who can be trusted. Frequently one doesn't even have provenance information for a media item, which is a prerequisite for making a trustworthiness judgment call.

The "hocus pocus" is literally the point that distinguishes this scheme from the plainly non-working system that we have today.

Just because something is different from something that doesn't work doesn't mean the different thing will work: most things don't work.

I'm not seeing anything in your description that couldn't also be said about using PKI to allow people to sign stuff they published. (And better: it's not broken by non-deceptive edits, like cropping or brightness adjustments)-- except that PKI already exists so we can't imagine that it would magically solve these problems. :) (and there is a long long history of people thinking varrious PKI ideas will magically solve varrious problems and then failing to solve them at all)

(I mean cameras that produced 'signed' output is also a thing that has previously existed too: https://blog.elcomsoft.com/2011/04/nikon-image-authenticatio... and was never particularly popular even with the target market)

Trusted by whom?

Trusted by whatever certification entities (ideally multiple ones, from non-aligned states) decide to deem a certain piece of hardware as "trusted". And then by extension, by any individual that trusts those particular certification authorities.

Yes, I'm aware this leaves the door open for subversion of certification authorities by nation-states. It's still better than the nothing we have today.

Your thoughts are very interesting, but I'm afraid remote attestation may not be enough. Processor instructions aren't sufficient as verification. Suppose I use AI to create a deepfake, then open up my phone, grab the wire which goes from the camera to the microprocessor, and rewire it so it goes from an Arduino to the microprocessor. Then I program the Arduino to send a signal to the microprocessor just like the one my camera would send to the microprocessor, had the camera just taken my deepfake as a photo.

I suppose a chip inside of the camera (in a tamper-resistant housing that self-destructs if you attempt disassembly) which signs the photo before sending it to the microprocessor could be sufficient in practice. Especially if the provider of the key infrastructure embeds a separate key in every camera, and monitors for abuse.

Anyway who is gonna build this? Perhaps Google could start selling special cameras to journalists, such that if the photos in a story on Google News were taken using that camera, the story in Google News gets a special checkmark next to it?

It wasn't elaborated on in the original comment, but yes you're absolutely correct, the imaging sensor itself would also need to be part of the cryptographic chain-of-trust. This is likely the most crucial part of such a system, as all the other hardware aspects are already starting to become generally available.

> Anyway who is gonna build this?

If and when specialized imaging sensors become available, I don't see this being a huge lift. One wouldn't even need purpose-built cameras for all but the most high-assurance recording requirements. Remote attestation and trusted code execution is already being incorporated into most newer devices. Bringing it all together would mostly be a software problem at that point.

Of course, there would also need to be a value proposition to justify any kind of work towards this. I personally see this as an inevitability given the frightening pace of progress in the ML field, and the increasing public concern over things like deepfakes.

Thank you for your comments!

You could start a startup building the specialized imaging sensors, with the aim of being acquired by $bigtech so they can tell congress "yes we are doing something about deepfakes" and eventually tell consumers "you can tell that photos taken by our devices/posted to our site are real, the other sites are full of BS"

I don't see how it all comes together without a partnership between the tech companies that display media (Youtube, Facebook, Twitter, etc.) and the tech companies that provide hardware (Apple, Google, Samsung, etc.) Maybe what's needed is to go through a nonprofit like the Partnership on AI.

Google is an interesting target because they are basically the only company that's both a media displayer and a hardware provider. If Android devices get a "verified non-deepfake" capability before Apple devices get it, I could see that as being a significant leg up for Android actually. It encroaches on Apple's traditional branding of "we make products for important people". I wonder if the Titan M has capabilities which could be leveraged. https://android-developers.googleblog.com/2018/10/building-t...

Google would probably sell loads of Pixel phones if it upranked "verified non-deepfake" Youtube videos recorded using Pixel cameras & microphones. Every aspiring Youtube star would know to record using a Pixel. The deepfake thing is a pretty good excuse to defend against an antitrust lawsuit, especially if Google can say that the capability will eventually roll out to competing hardware.

Starling Lab focuses on data integrity for video evidence. It’s worth checking out their recent efforts here: https://sfi.usc.edu/news/2022/06/33571-starling-lab-and-hala...

Your comment inspired me to post a blog post on this very topic. You can find it here:

https://news.ycombinator.com/item?id=33558566

Cryptography. Honestly, all sources (news outlets, self publishing celebrities, etc.) should start cryptographically signing their content one way or another. Including video and audio. (I haven't looked up whether there are existing, easy to use/practical schemes for this, but I guess there are and no need to reinvent the wheel. Even if the actual tools are missing.)

>Multiple camera angles. A simultaneous left / front / right view of a person speaking

With enough training data and modified network this can also be done.

Reply on news.ycombinator.com