Six principles, generative AI harms, Guardrails, prompt shields, transparency notes.
Choose a difficulty tier. Easy tests recall, Medium tests scenarios, Hard tests distractor traps. Questions in each run are shuffled.