Prompt-Stealing-Attack-Model Evaluating Prompt Stealing Attacks in AI-Driven Assessment Content Generation