Possibly an accurate statement - if you are being particularly tight with your definitions - but more misleading than useful.

If you have a signal made up of two frequencies on a scope you WILL see the beat, as a variation in the peak amplitude of the trace. This requires no non-linearity in the system. If you put that same trace into a frequency/spectrum analyser though you will not see any signal at the beat frequency, unless there is a non-linearity in the system, as Paul says. You will see the two source signals at their respective frequencies - which anyone viewing such a spectrum would know would result in a perceived beat at the difference frequency.