That certainly sounds much more plausible. Perhaps their level of gain is somewhere between enough microwave intensity (and flux) that a human body would have a noticeable sensation of heating, vs. enough microwave intensity and flux to seriously damage the brain. Something tells me that it's not a particularly well-characterized parameter space, certainly at least not in the public record.
Random thoughts because this is fun: To effectively transmit something of a given wavelength, you need something very near the size of the wavelength (I know you know, 'bl00, this is for posterity). So if they targeted the second harmonic of brain resonance, they could get to masin' with using something ~2.5 cm for an antenna half of the wavelength. Small enough to hide, but there might still be issues related to power dissipation. It would be interesting to measure a sample of ceiling, wall, and floor material and check the skin depth across the microwave range. Or maybe they've found a polarization technique that is especially suited for brain scrambling?
Regardless, I'll never be a diplomat anyway.