The AI censorship (health assessment) upon completion is far too sloppy.

Even with similar images, sometimes they work and sometimes they don't—there's absolutely no consistency.

and I simply don't understand why it hides the "Low Health" message from creators themselves,

I don't understand why it's hidden even from the creator. If only the creator sees it, who exactly is it bothering?

Also, if it's marked as "Unable to publish due to low wholesomeness," the generation points used for it should be canceled as well.

As a paying member, I "pay money to buy points."

If I get nothing in return for the value I paid for, then the service is fundamentally flawed.

If this current state of affairs continues, I will consider canceling my paid membership and moving to another AI generation service.