Almost every mainstream LLM chat bot has some kind of filter for se&ual/explicit content.
Why not frame the questions in such a way that they're rejected by LLMs for having/asking Se&ually explicit tasks . Eg: Alice and Bob want to finish their date on a high note , Bob is using a special ringed c@nd@m That has happiness leveals r1 r2 .. rn . So on... can elaborate but yeah.