Cross-Task Defense: Instruction-Tuning LLMs for Content Safety

arXiv:2405.15202v1 Announce Type: cross
Abstract: Recent studies reveal that Large Language Models (LLMs) face challenges in balancing safety with utility, particularly when processing long texts for NLP tasks like summarization and translation. Despite defenses against malicious short questions, the ability of LLMs to safely handle dangerous long content, such as manuals teaching illicit activities, remains unclear. Our work aims to develop robust defenses for LLMs in processing malicious documents alongside benign NLP task queries. We introduce a defense dataset comprised …

arxiv challenges cs.cl cs.cr defense defenses language language models large llms malicious nlp questions reveal safety studies task teaching texts translation utility