The AI safeguards are indeed a joke, you can get around their classifier by simply masking out all the unsafe words and it will happily work on your rootkit.