My take on these situations is that they are NP-complete in a way: looking at something (the assistant voice for example) with a clear and objective mind and trying to figure out a way it's offensive is much harder that verifying whether an offensive attitude is being portrayed by that thing.