it's widely accepted that ultrasound applied to the skull is even audible, this was discovered shortly after WWII: Gavreau in 1948 and then Pumphrey in 1950, who proposed the middle ear ossicles simply form a low-pass filter, although the situation is a lot more cloudy to this day: the topotonic relationship or the subjective frequencies don't seem to correspond to their actual frequencies. It's unclear if that is due to intermodulation distortion (non-linear frequency mixing) or perhaps even simply the frequency relationships being novel to the untrained ear (for the first time exposed to ultrasound, well not really, when biting on food or generally clenching the teeth and associated friction between upper and lower teeth are bound to emit countless ultrasounds, but not exactly harmonic oscillators, so perhaps the brain has only learn't time domain associations and little to no frequency domain associations for these higher frequencies)

If you believe ultrasound categorically can not penetrate the skull, while the much longer used Weber and Rinne tests clearly demonstrate bone conduction of sounds by placing a tuning fork against the skull, so I wonder how you explain that: audible frequencies can pass but independently defined ultrasound is unable to pass?