From what I can see current mm wave devices are using the 3GHz-50GHz region. Airport scanners are 24-30GHz. I found a paper that used a 140GHz (0.14THz) light source (above 300GHz is far-infrared) to test the properties of silk.
Sub-Millimeter-Wave Transmission of Silk Fabric
https://file.scirp.org/pdf/JCC_2017122809425988.pdf
There is some information here:
https://www.polymtl.ca/phys/photonics/papers/2014_AOM_Silk_foam_THz_waveguides.pdf
To make this useful we need to compare it with other fabrics like cotton. Here is a paper on the transparency (I think) of different fabrics in the THz range.
https://www.researchgate.net/public...omain_spectroscopy_for_textile_identification
We know that silk passes light through, so it makes sense to see that it is actually more permeable as the frequency approaches the visible light range. The region of interest (<1THz) is crowded to the left side of the charts but here silk also seems to transmit more than other fabrics. It is important to note that early in this paper they mention that they have to dry out the samples to make sure the results are accurate. This could mean 2 possibilities: 1: moisture dominates the transparency of silk and other fabrics in a certain range. 2: these results may not be relevant to the transparency of silk as worn with the body's natural humidity.
The messages here are mixed, but there is no certain conclusion that silk blocks millimeter waves, especially as they approach visible light. It's hard to find information on the 3GHz-100GHz range which 5G technology is expanding into, but we have boxed it in on both sides as not being a likely shield. If it were to be a good shield in this region, it would have to mean that it becomes reflective or absorptive within one decade of frequency and then flips back again before 100GHz. This would be abnormal behavior.
It seems possible to me that as silk is birefringent, it could mitigate mm wave severity by scattering it's polarization. We know now that being highly polarized is one reason manmade EMF is harmful. However it would be good to confirm the birefringence at the frequencies of interest.