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.”
Problem is that broadly speaking, you would only sign the stuff you want to sign.
Imagine you had a president that slapped a toddler, and there was a phone video of it from the parents. The white house isn’t about to sign that video, because why would they want to? Should the journalists discard it because it doesn’t carry the official White House blessing?
It would limit the ability for someone to deep fake an official edit of a press briefing, but again, what if he says something damning, and the ‘official’ footage edits it out, would the press discard their own recordings because they can’t get it signed, and therefore not credible?
That’s the fundamental challenge in this sort of proposal, it only allows people to endorse what they would have wanted to endorse in the first place, and offers no mechanism to prove/disprove third party sources that are the only ones likely to carry negative impressions.
But then the journalists have to check if the source is trustworthy, as usual. Then they can add their own signature to help other papers check it
To that extent, we already have that.
I go to ‘https://cnn.com’, I have cryptographic verification that cnn attests to the videos served there. I go to youtube, and I have assurances that the uploader is authenticated as the name I see in the description.
If I see a repost of material claimed to be from a reliable source, I can go chase that down if I care (and I often do).
It’s not a challenge, because this is only valid for photos and videos distributed by the White House, which they already wouldn’t do.
The challenge is that it would have to leave out all the photos and videos taken by journalists and spectators. That’s where the possible baby slapping would come out, and we would still have no idea whether to trust it
