This ban probably comes as a response to this activist project by the Peng! collective: https://pen.gg/campaign/mask-id-2/ where they used morphed portraits for passports.
So if I understand correctly, the kind of attack enabled by this manipulation is to provide a way for non-passport holders to pass custom controls by using the legitimate passport of the person they morphed their picture with. Is there any other useful kind of attack possible with this scheme?
>Such manipulation of photos is typically invisible to the human eye, the researchers found.
I think its more like attacking a machine learning model - the difference in the photos may not be apparent at all, so a human reviewer might not catch something.
That is unlikely. Germany can't get enough from "non-integrating gap countries"-immigrants. They have the reputation of integrating really well into western society.