[sircastor]

As a random suggestion for a project to pursue, over the past few years I've thought that there needs to be some way to cryptographically sign media that was recorded by a camera - basically a way to say "Yes, this is a real thing that happened, and you can prove it with this signature." Especially given the leaps in audio, video, deepfakes, etc, we're all going to need a means for verifying that something is real.

I'll grant that something like this is not a panacea - we need good solid media literacy. But I think a technical solution will help us out.

[wiihack]

Something like thats exists already! The organisation behind it is called C2PA :) https://c2pa.org/

[obnauticus]

I’ve worked on this specific problem you are describing but for other reasons.

The issue will be once someone breaks the secure enclave in the CCD, all of the video they produce will now be trustworthy. CCDs are pretty lightweight in general as most of the heavy lifting is meant to be offloaded to the ISP—so the on-die processing for the CCD is generally very lightweight and wouldn’t have buffers large enough to sign each frame. Also there is no standardized way of creating a secured MIPI connection. You can do some clever things to mitigate for this, but it is far from what I’ll call “best practice” and more along the lines of a bespoke solution.

[badpun]

Sounds interesting! It would probably require a hardware solution though - a private key and encryption chip tightly coupled with the rest of the camera hardware, so that it cannot be reached from the outside (to extract the private key).

[bjornsing]

Security by obscurity... But maybe good enough for this use-case.

[aardvark179]

Such cameras already exist.