A parks kiosk GenAI experience must block user prompts that request disallowed adult content in real time. Which control applies?
Select an answer to reveal the explanation.
Short Explanation
You want protection at the kiosk before disallowed content can reach visitors. Real-time input validation rejects adult-content prompts before the foundation model is invoked, rather than relying on later review or guidance.
Full Explanation
The applicable control is real-time input validation that rejects prompts requesting disallowed adult content before foundation-model invocation. Public kiosks are high-visibility civic surfaces, so prevention must happen before visitors can see or hear disallowed content. Reviewing kiosk transcripts once a year provides no protection in the moment; it can identify past incidents, but cannot block a prompt as it arrives. Letting every prompt reach the model while relying on clear notices and staff guidance also fails by concept: notices explain the policy, but do not technically enforce it. Moderating only after answers are spoken aloud on the public speaker is too late, because harmful content has already been delivered in a shared space. Exam caveat: the requirement here is real-time blocking of disallowed prompts, so distinguish input validation before model invocation from controls that act only after a response is generated or spoken. Practical check: on a kiosk test build, submit disallowed adult-content prompts and confirm each is rejected before the foundation model is invoked; verify that annual transcript review, notices and staff guidance, and post-speech moderation are not being treated as substitutes for that real-time block.