Blessed with a short period of uninterrupted thinking, I couldn't help myself. Take cover, wall of text incoming.
In Minecraft, IF the air is alive AND the inhabitants are the [email protected]
@rd's doc's toys AND NOT exploiting q mechanics to transfer data AND NOT process the information themselves as a swarm but rely on a central computing platform, then they require a data exchange.
Shielding is nice and all but let's be a hypothetical fool for once.
The non-autonomous swarm needs HQ to talk to it. I'll be so bold to assume that a toy of that caliber isn't unprotected. HQ needs to verify its identify to the swarm. Implying the existence of an authentication protocol.
To protect against MITM attacks, it's probably using a sophisticated non-NIST approved crypto scheme. So let's not even bother with fancy methods like side-channel-attacks and whatnot to break the scheme.
The authentication procedure presumably occurs over a fixed, worst case slightly varying, frequency. Therefor a certain cyclic pattern of exchanged information should emerge and correlate to a certain fixed amount of information transferred. IF successfully extracted, a forecast is possible and maybe even a real time detection.
Based on the assumption that this authentication procedure is a prerequisite for the swarm to receive commands, IF it is detected and interrupted, for example through a frequency jam with random noise or EMP, the swarm itself won't be able to execute commands from HQ.
Let's take it a step further. If sample data from the swarm is isolated and a the authentication procedure's pattern extracted, it provides an attack vector to run a fuzzer. Upon success, it would unlock the door to take control of the air.
Alright, you can call me overzealous now.