Fetching the paper…

RapGuard: Safeguarding Multimodal Large Language Models via Rationale-aware Defensive Prompting · Around