There does not seem to be a way to report when the generation is clearly misunderstanding the prompt, and trying to ask it for clarification is even less helpful.

I provided an initial prompt, which did not meet my intent, which is fine.  But after multiple clarifying prompts, the generated image ignored the clear issues identified (it turned a woman into a man) and did not give any option to flag it as inaccurate.