When I generate content on most AI's including Grok, I ask it to fashion a prompt first of the subject I want and ask it to make sure that it does not violate a

When I generate content on most AI's including Grok, I ask it to fashion a prompt first of the subject I want and ask it to make sure that it does not violate any TOS or CSAM policies. I also instruct it that the prompt should be usable by most AIs. It fashions the prompt. When I use the prompt, the system complains that the prompt violates the TOS. I then ask the AI to locate the troubling aspect of the prompt. It says that it has and provides an alternative, safer prompt. More often than not, this newer prompt is also flagged as inappropriate. This is very frustrating even when the original intent is not to create content that violates any public AI policy. From my experience, both users and the technology make mistakes.