I know. It's like we are going to need reverse Zero Days after a while. As soon as they publish it becomes the old new way to detect Deepfakes.

-1 days

Yes, but if the "fake detection" technique is computationally much faster than the generation technique, having access to a better oracle is not so helpful to the deepfake creators, who are bottlenecked by the other phase.

I would interested in a way to automatically edit real photos of people in such a way that it looks the same but now tests positive for being a deepfake.

Especially if it’s differentiable

At some point in the future, AI will make better humans than naturally occurring humans.

The built in "touch up my appearance" feature in my phone, various video chat software etc. means that future is already here [1].

[1] https://mothership.sg/2021/03/japanese-biker-actually-man/

Tangentially reminds me of the film Surrogates, where people only went out in their surrogate, a remote controlled android - in some cases, a younger, better looking version of people. In other cases something completely different of course.

Friedrich Nietzsche called that "Superhuman", or "Übermensch".

And that we're just a tightrope stretched over the abyss from the monkey until ai.

From where or from whom will 'ai' be getting its moral outlook on the actions it takes?

From its arbitrary mesa-optimisation goal, of course.