Analyze picture content, not picture keywords

I am against censorship. There is no proof whatever that “morally wrong” fantasies cause any crimes or problems. We should be able to post and share any picture with those who have “Mature mode” on. However, if you need to comply with Chinese regulations, you can at least configure your post analysis to look at picture content instead the words that were used to generate it. I have many examples I could give. For some task results I could naughty words, but the pictures that resulted are perfectly wholesome. There is nothing in them that anyone could object to. Yet I am prevented from posting them with an accusation of “sensitive content.”