Public/Private key cryptography will be needed to curb a lot of the AI trickery that is coming, I think.
There's not a realistic expectation of slowing down the progress of AI image generation's realism, so we will need to adopt methods of verification that even AI cannot duplicate and assume that all images are fakes unless signed by a verified private key.
Difficult question. If an AI will successfully pass a Turing test, then this AI can trick the whole PKI mechanism (even if quantum computers exist). The only thing when the AI will fail the mandatory appearence personally - "in the flesh".
But the KYC will not be enough, this is not a question.
On the other hand there will be some specialized AI system that could filter the " tricky" AI.
And there will be AI that tricks the filtering AI and so on.
