They say that it are 2 square grids seperated by at thin film, with the grids rotated over 45 degrees. That would make it a cage in a cage.
Difracting the signals will only bend the path of the signal. So the signals just end up in another place and that also depends on the incoming path of the signal. To 'absorb' this signal they have to be able to bend al signals, regardless of ther original path/direction, to a known absorber. The 'diffraction grid' has to be so tight that there are no gaps bigger than the wavelength they want to filter or there will be leaks. But they still needs some gaps somehow that are big enough to allow the longerwanted wavelength signals to get through the grid.