They don't intend to dictate who can authorize media, only provide a verification mechanism that the media was sourced from the place it claims to have been sourced from and is unaltered.
I think of it as https but for media content.
It seems like they're on the right track. I think the key is to keep scope creep to a minimum. As soon as someone tries to add DRM, for instance, the whole effort will go up in flames.
The inverse is equally problematic and harder to solve: those in power discrediting real photos/videos/phone-calls as “deep fakes.”
Not releasing AI models doesn’t stop this. The technology being possible is sufficient for its use in discrediting evidence.
Signing real footage isn’t sufficient. You can get G7 to sign an official conference recording, but could you get someone to sign the recording of them taking a bribe?
Generating deep fakes that hold up to intense scrutiny doesn’t appear to be technically feasible with anything available to the public today. But that isn’t necessary to discredit real footage as a deep fake. It being feasible that nation state level funding could have secretly developed this tech is sufficient. It seems we are quickly approaching that point, if not already past it.
This is true. Although witnesses and/or the person behind the camera could sign the video. They might want to, in fact, if they thought they might be witnessing something illegal and might need to defend themselves later.
I guess I'm imagining a future where signed videos are common. Unsigned content or content signed by some random entity would draw suspicion. Maybe not enough to keep people from seeing it, but enough that it wouldn't spread like wildfire like it does today.
There could be disputed videos, too-- Where one party signs and one doesn't. Or maybe a situation where two parties secretly ally against another to run a more convincing smear. Hmm, there's all kinds of weirdness that a system like this might create. Maybe a cyberpunk author could explore it further :)
Combining all of those things would make it impractically difficult to fake a scene without knowing what you want to fake in advance as well as developing credible witness reputations even further in advance.
For example, imagine a car accident caught by dashcams. You'd not only have your own dashcam footage certified to have been produced no later than the event by a timestamping service, but also corroborating footage from all other nearby traffic also certified in the same way but by other, competing services.
It'd be the future equivalent of having many independent witnesses to some event.
Maybe it won't be necessary to go quite as far, but I think it would be possible for recordings to remain credible in this way, should the need arise.
The idea of verifying media has yet to permeate the edges of mainstream thought. People are arbitrarily credulous or dismissive, based mainly on who's sharing the video.
It wouldn't solve any of the fundamental problems of trust, of course (namely, the issue of people cargo-culting a specific point of view and only trusting the people that reinforce it). But, it would at least allow people to opt out of sketchy "unsigned" videos showing up on their feeds.
I guess it would also allow people to get out of embarrassing situations by refusing to sign. But, maybe that's a good thing? We already have too much "gotchya" stuff that doesn't advance the discourse.