Playing whack-a-mole with technologies that have fundamental risks baked into the system itself seems like a losing game. We're inventing new technologies simply because we can, ignoring and papering over the social impact, then chasing even more complex solutions to the new problems created.

Technical solution to a social problem. It's not that it's impossible to tell the difference between real and fake, it's that there's no consequences for posting fakes.

(A "solution" which requires end users to authenticate everything themselves is going to have the same level of adoption as pgp, i.e. near zero)

I like a green checkmark next to a photo better than giving more power to whoever you think should be the one determining and enforcing those consequences.

The consequences I mean are those of the public: proving that the ruling party forged images isn't going to get people to change their votes. You can build the green checkmark, but you can't get people to care about it.

I see this as completely different issues. Whether AI exists or not this is and was a problem. I’m proposing technical solution because AI fakes are a technical problem. And yeah if people get used to green checkmark you can’t just pump fakes willy nilly, you actually have to put some effort, as before AI. I guess the problem I want to solve is having to live in a world where you have to assume absolutely everything is 100% fake all the time. Because this just puts everyone in their respective truth bubbles even more. Any objectivity we can return to the world is a good thing, IMO.

First you have to solve the collect action problem of getting everyone to care about the green checkmark, then you have to solve the new technical problem of just making photographs of high-quality altered images.

I fully acknowledge the second issue, and have offered some potential solutions in my other comments, but I think this can have tremendous impact even if you don’t solve the second issue in first draft.

It's not even a solution because it completely ignores the analog hole.

I explicitly acknowledge analog hole in my other comments in this thread and suggest advanced sensors (as used in existing biometric authentication) and LIDAR as potential solution. Nevertheless, I think this system can be very useful even without those countermeasures and even if it can still be bypassed with effort. Additionally, if the signatures are securely time-stamped it makes proving which photo is genuine straightforward.

It's not even a solution because it completely ignores the analog hole.

Not to mention the inevitable abuse by whoever controls the keys.

I take an authentic image with a camera, crop out the part of it that it's not aligned with my goals, put a reference to the authentic image. How do you get the authentic image to compare it with my edit?

And how do you solve the problem of a powerful entity telling you that the cropped image is authentic, or there will be consequences? People know what's going on, they bend for fear of breaking. No technology is changing that.

You have to provide authentic image with your image for it to validate, just as you have to supply any intermediate certificates in TLS. If you want to crop to hide information you have to do this while taking the photo in the camera app.

Powerful entity can tell you to install their root certificate just like it happens in China. This is a different problem.

If browsers require authenticated images as suggested, then CSS will probably grow the ability to apply a masking layer to crop, filter, transform the image.

Advertisers are going to want to postprocess images and have the green check.

Well I don’t think all images would have to be authenticated, just photos you take on social media and dating apps also seem like a good potential use case, if you want to prove authenticity.

You could of course fake green checkmark but this is already an issue with faking browser UI elements, including address bar–it’s a known problem.

One way to solve this is having a browser picker, that is a button you click in trusted area of browser UI that lets you inspect any image on the page. Similar thing is possible with smartphone UI, you go to app overview/carousel and long press on an image in app this launches photo inspection portal.

Just to provide some context : The reason people believe this particular doctored (or AI enhanced) image "could be true" is because this "toolkit approach" has been employed by the Left Wing in India. The Right Wing in India regularly points it out on Twitter. So if both sides are slinging mud, its race to the bottom for the getting lowest common denominator on their side. I sometimes wish why can't we unbundle these left vs right issues and decide stuff case by case. Hence I support TruthGPT by Elon.

> I sometimes wish why can't we unbundle these left vs right issues and decide stuff case by case.

That's the opposite of what we should do. If you do case by case then people will always find reasons why it's no big deal if "their" side did it while it's really bad when the other side did it.

No, we need to universally condemn this. No matter who did it and when and how. This is lying and propaganda, and it's bad even if I agree with the ends. But the means are to be condemned by everybody.

Kind of the dual to "I disagree with what you are saying but I'll fight for your right to say it." It's now "I agree with what you are trying to achieve but I'll fight against the way in which you are trying to achieve it."

You can't unbundle a person or partially punish a person. The people in power won't agree to such a situation. These mistakes happen collectively within the institution. I think the only way out would be to start with a moral clean slate for everyone like the debt cancellation e.g. Biblical Jublee [0] [1]

[0] https://en.wikipedia.org/wiki/Jubilee_(biblical)

[1] https://youtu.be/CZIINXhGDcs?t=736

What if I just reframe my picture instead of cropping? Every photo is a choice. A choice of what to show and what not to show.

I think you’re missing the point, this is not a technology to create politically neutral photos. The way you take a photo and even the way you edit it later will all be your choice. It’s also your choice what to take photos of or not in the first place. All this does is it proves the photo was created by

(1) a trusted application

(2) using a builtin device camera

(3) of a device running attested trusted operating system

(4) and any subsequent edits reference such photo’s original hash.

Hopefully stronger assertions could come in the future. This is just to combat digital forgeries and synthetically generated images. It also does nothing against real-world forgeries like actors wearing costumes or makeup.

> I take an authentic image with a camera, crop out the part of it that it's not aligned with my goals

If it is cropped, then it would not be provably authentic. So anyone seeing it, could see that it is modified.

> how do you solve the problem of a powerful entity telling you that the cropped image is authentic

Well, because you can check. You verify the image yourself, see that it is not authenticated, and then you know.

> You suggest giving up.

Not at all. I'm not sure what I said that makes you think that.

What I say, though, is that we don't know (either individually or as a society) how to operate in this new environment. It's making it hard to stay sane as an individual, and to stay functioning as a society.

Maybe we don't know how do operate in this environment, because we don't really know how to collectively operate in any environment?

Humanity is, after all, crawling out of at least the last 3000 years of recorded history of rulers telling everyone else how to live, like it or not.

We've only really been doing this democracy thing semi-successfully for a handful of centuries.

In the overall scheme of things, we're still pretty new at not being ordered around by people stronger than us. It's a hard habit to break on both sides.

I have read your comment as we just have to accept this new environment and just ‘cope’ with it, whereas my feeling is we don’t have to accept it at all. I don’t want to cope with AI and question if what I see everyday is in fact real. Instead, anything produced by AI should have an asterisk next to it. But this is impossible to enforce, so verifying what’s real by default is the next best thing.

How will this protect (the first "workaround" that popped into my head) against modifying/AI generating an image, printing it on high-quality printer and taking a photo of that image in good (or maybe bad?) lighting?

You can harden against it using stereo cameras (3D) and different type of cameras, e.g. IR, verify with for example LIDAR, since some new premium smartphones are getting it already, this will tell you if you’re looking at a projection. Also see what smartphones are already doing for face authentication tech. Now, that said, you can never defend against it 100%. The most obvious is you can always fallback to just hiring actors/makeup/etc. If someone is determined enough (major actor) they can probably still produce fake pictures and authenticate them, but it would be significantly harder, and basically impossible for majority of people.

Photos have authenticated timestamps.

Older photo is the real one.

Geez. Techies trying to come up with tech solutions to fundamentally non-tech problems. The photogtapher can just zoom in or get closer to the object to crop the image before it's even taken.

Oh you now want a 360 degree shot to prove this didn't happen? Well then they stand behind a tree or a car or some other object to hide parts of the scene behind. etc. etc. There is no technical solution to this problem.

It makes the attack in article not possible and eliminates common fakes in existence right now. Perfect is the enemy of good.

At huge cost and essentially nonexistent benefit I'd say that crappy is also an enemy of good.