I don't know what type of filtering is applied to detect radio-transmissions. If the detection equipment is not looking for microwave-attacks, they could indeed get filtered out as noise. But again, at intensities that induce the Frey effect you illuminate a neighborhood well above the normal noise-level. You can't hide this.

I disagree about the shielding of microwaves because the shielding is obviously very effective. If it weren't so, we'd feel physical discomfort with our face less than a meter from the magnetron.

For the properites of shielded windows, I refer to "Shielding Effectiveness and HPM Vulnerability of Energy-saving Windows and Window Panes"[0]. The three panes they measured range from 15 dB to 35 dB attenuation. Now scaling an attack by 15 dB is not generally easy. It may mean you have to reduce the distance and operate from a van instead of a flat for example. By 35 dB we're talking about three orders of magnitude which makes scaling impossible because it'd cook the whole neighborhood.

[0] https://www.researchgate.net/publication/324950137_Shielding...

Your microwave oven has a mesh in the door that you look through all the time, it keeps the microwaves inside, and keeps you safe. This is a well known technology.