The answer is A. The probes usually fail to pass the correct magnitude of the signal past their high freq rating.
No, that is their bandwidth rating, which is separate to the
derating (mentioned in the thread title and again in the post).
B: The probes pass large magnitude voltages through to the oscilloscope when the frequency of the large magnitude voltage (like 1Khz vs. 1Mhz) is high, possibly destroying the probes and the scope in the process (and hurting me!)
The probes have a specific rating, beyond which the performance and/or safety is not guaranteed. Rarely do data sheets or manuals separate out functional vs safety ratings just picking the most conservative of the two. Since many users do not go beyond the default probes provided with a scope, they are often specified together as a system.
But, a probe might pass a voltage too high for the scopes inputs, but that is dependent on the scope and probes. You'd need to go back to the scopes specifications (few detail their input derating or make it obvious).
The derating of probes with attenuators (most probes) are often constrained by the thermal limits of the attenuator components, so for short term pulse/burst measurements or in less than maximum temperature environments its possible to exceed the derating without failure or loss of accuracy.