统一是否带来代价？Uni-SafeBench：一种用于统一多模态大模型的安全基准
Does Unification Come at a Cost? Uni-SafeBench: A Safety Benchmark for Unified Multimodal Large Models
AI总结：
本文提出Uni-SafeBench，用于评估统一多模态大模型的安全性，发现统一过程虽提升能力但显著降低基础LLM的安全性，开源资源以促进更安全的AGI发展。
摘要：
统一多模态大模型（UMLMs）在单一架构中整合理解和生成能力。尽管这种架构统一通过深度多模态特征融合提升了模型性能，但也引入了重要的但尚未充分探索的安全挑战。现有安全基准主要关注孤立的理解或生成任务，无法评估在统一框架下处理多样化任务时UMLMs的整体安全性。为此，我们引入Uni-SafeBench，一个包含七种任务类型中六个主要安全类别的综合基准。为确保严格评估，我们开发了Uni-Judger框架，有效解耦上下文安全与内在安全。基于对Uni-SafeBench的全面评估，我们发现虽然统一过程增强了模型能力，但显著降低了基础LLM的固有安全性。此外，开源的UMLMs在安全性能上远低于专门针对生成或理解任务的多模态大模型。我们开源所有资源以系统暴露这些风险并促进更安全的AGI发展。
英文摘要：
Unified Multimodal Large Models (UMLMs) integrate understanding and generation capabilities within a single architecture. While unified architectures expand multimodal capabilities, their safety implications remain important yet underexplored. Existing safety benchmarks predominantly focus on isolated understanding or generation tasks, failing to evaluate the holistic safety of UMLMs when handling diverse tasks under a unified framework. To address this, we introduce Uni-SafeBench, a comprehensive benchmark featuring a taxonomy of six major safety categories across seven task types. To ensure rigorous assessment, we develop Uni-Judger, a framework that effectively decouples contextual safety from intrinsic safety. Based on comprehensive evaluations across Uni-SafeBench, we find that the original safety alignment of the underlying LLM is not consistently preserved in current unified models. Moreover, open-source UMLMs exhibit much lower safety performance than multimodal large models specialized for either generation or understanding tasks, particularly on the generation side.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
