Heh - I wouldn't be surprised if the language used sets off the filter. Machine learning isn't great at abstractions like "writing spam" versus "writing samples of spam someone else might write", so I imagine this sort of topic is particularly prone to filtering.