Pangram agrees with you. About 25% of the text trips the detection threshold, mostly towards the later half.

I don't want to make any accusations, just give some evidence to the above comment.