The previous post got me to try my hand at some Science®. Please note this knowledge applies only to the Bluesky censor-bot, your robot's arousal level may vary.
So, the following very scientific picture was auto-tagged as "Non-sexual Nudity":
This second experiment was also picked up by the purity-bot:
While I theorized that perhaps it was the raunchy rectangles that were being flagged, this proved to be incorrect as the next tests was negative:
My suspicion that the robot was looking for Vagina Bones also turned out to be erroneous:
However! When knees were re-introduced into the test, suddenly, the machine manifested and tagged the image as Non-sexual Nudity once more:
A workable theory began to take form, as censoring the knees did not wake the nudist-huntbot, suggesting that perhaps its nakedness sensor may be patella-based:
In further testing this theory I realized I'd unknowingly introduced too many variables (the previous test had sparkles! Sparkles!):
This came up negative, but I cannot be certain as to why. At this point it was clear my scientific method was lacking and I do not have the chops/patience for true evidence-based sciencing.
Before abandoning this study that would most certainly not pass peer-review, I conducted one final test:
Yes, it appears drawings with the evil hand do not trigger the abstinent algorithm. I believe artwork will be safe from automated suppression as long as only the evil hand is involved, though more rigorous testing is needed to be absolutely sure.
Anyway science is hard and robots are basically all science, even the ones who do nothing but look for potential nipples, so we should fear their inscrutable ways because you never know what other body parts they may take scientific offense to.
