Hacker News Reader

Comment by arghnoname | Hacker News Reader

While these images are stylized and would never be interpreted as real, we are just rapidly entering a world where photos should not be trusted. Or at least they shouldn't be trusted anymore than a sketch.

It's not the end of the world--no one gets confused by a political cartoon showing political opponent X dressed as legacy enemy Y (e.g., Trump as Hitler). I think the fuss about safety here is overblown once we more widely accept that a 'picture perfect' rendering of an event is as unreliable as a sketch of an event.

It may even be a positive: it will give us all plausible deniability if real but embarrassing photos get released.

View on HN

> It may even be a positive: it will give us all plausible deniability if real but embarrassing photos get released.

Honestly i don't think it will. Rather it will make that problem worse. Only when everyone has nudes will it feel less embarrassing to see your AI-self in explicit situations being passed around.

Basically it'll have to be so common that people don't care to pass it around.

I thought Photoshop already did that.

Even before Photoshop, photographs could be altered and airbrushed. Even so, faking a convincingly realistic image in Photoshop takes effort. Generating a plausibly realistic image using Stable Diffusion requires almost no effort whatsoever, just access to a GPU. This is a significant change.

"we are just rapidly entering a world where photos should not be trusted"

That train departed years ago.

If push comes to shove, we already have all the cryptographic primitives needed to produce and disseminate trustable imagery.

However, society at large doesn't actually care.

I don't think that cryptography can solve the problem of trustable imagery. What sort of system can do that?

The Camera manufacturers are putting DRM and TPMs on camera to sign the images.

It will be broken the minute the launch, but that won’t stop them from trying.

Cryptography lets you certify attestation. So an organization can certify that an image is legitimate, and you can cryptographically verify that this organization has indeed certified that this is a legit image.

That's it. It doesn't verify that the image itself is real, only that some organization has put their stamp of approval on the image. You can verify that the image wasn't tampered with between when they approved of it and it got to you, but you can't verify that the image was real to begin with.

On principle, you could build a chain of certification into the camera itself but this strikes me as a losing battle because you could just stage whatever you want in front of the camera.

The impetus then is on organizations to build that trust.

My question is how would those stamps exist in the first place? Is the idea that Canon or w/e will ship their physical cameras with keys, and sign those images with the keys. Now when you go to look at an image it'll be verified Canon, or w/e.

In that world wouldn't keys leak pretty easily? The key exists on the device. Is there a way this sort of stuff is actually viable? Or do i have the model entirely wrong?

You're right, hardware keys will leak. The idea is not to put trust into hardware, but to put trust into organizations.

Some photographer gives images to a news organization, and they take the photographer's word for it that these images are real before they sign and redistribute them. They trust the reporter, and you trust the organization. Or if the reporter has built enough credibility, they can vouch for the images themselves and you can trust them directly.

Cryptography allows you to verify that this reporter or organization attests that the images are what they claim. It doesn't allow you to verify whether the organization is worthy of your trust.

Either way, the system relies on being able to trust people, not things.

> The impetus then is on organizations to build that trust.

And that's where it all falls apart.

Yep cryptography solves trust, that’s why we don’t need CAs.

Reply on news.ycombinator.com