By invoking the inverse-sequare law, you are doing a couple things that are incorrect simultaneously. First, you are assuming a false equivalence between a beamforming antenna that sends concentrated, coherent beams of RF at a single target with an ordinary antenna that radiates energy isotropically. Second, you are also assuming that the power density of the system must be high in order to bring about any physiological effects. This is not necessarily true. It may be the case that self-assembling nanotransducers surreptitiously placed in the brain may only need to harvest nanowatts to induce profound changes in neuronal activity. The amount of energy that actually reaches them could be very small, and yet still achieve the desired effect.
LOL it's my fault now for "assuming"... how about you post something coherent beyond "may", "could", and vaguely relevant speculation. What is "high" power density? How much does your imaginary self-assembling 5G receiver need? How much does the brain-prodding thing need? How much power can the hypothetical magical physics-defying antenna harvest? Does this involve some energy storage for when I go to my basement and out of 5G signal range, or do I just snap out of the mind control thing then? "nanowatts" (better than -60 dBm) is extremely optimistic, in reality you'd be dealing with picowatts if that.
...
Just a little back of the envelope (or HP-48 emulator) scratching, with 2000 Kcal/day at 25% allocated to the brain, and about 100X10^9 neurons, the natural energy budget is about 0.000005 cal per neuron per day. With that they must both operate their normal celular biological process and generate electrical energy for signaling. Noting that a) the various talk, RFQ, etc were pretty much centered around a per-neuron interface, and b) a capacitor is probably one of the easier components to assemble, it seems like energy acquisition at sufficient power levels (voltages and currents) would be among the lesser of the problems. I would think it more likely to be done by taping biological structures which already maintain a potential (mitochondria, other neurons, etc) for the most part anyway.