跨语言跨模态检测有害提示，准确率提升超20%且速度更快。
OMNIGUARD: An Efficient Approach for AI Safety Moderation Across Languages and Modalities
- 利用多语言多模态对齐的模型内部表征构建通用分类器
- 多语言场景下准确率提升11.57%，图像提示提升20.44%
- 基于生成时嵌入复用，比最快基线快约120倍
大型语言模型（LLMs）的兴起引发了对其潜在滥用的担忧。当前主流防护方法是检测有害查询，但现有方法在低资源语言或非文本模态（如图像、音频）提示上易受攻击。为此，我们提出OmniGuard，一种跨语言、跨模态的有害提示检测方法。该方法首先识别在不同语言或模态间对齐的模型内部表示，进而构建语言无关或模态无关的有害提示分类器。在多语言设置下，相较最强基线，准确率提升11.57%；针对图像提示，提升达20.44%；在音频提示上达到新SOTA。OmniGuard通过复用生成阶段的嵌入，实现极高效运行（约比最快速基线快120倍）。代码与数据已公开于https://github.com/vsahil/OmniGuard。
原文摘要 · Abstract (English)
The emerging capabilities of large language models (LLMs) have sparked concerns about their immediate potential for harmful misuse. The core approach to mitigate these concerns is the detection of harmful queries to the model. Current detection approaches are fallible, and are particularly susceptible to attacks that exploit mismatched generalization of model capabilities (e.g., prompts in low-resource languages or prompts provided in non-text modalities such as image and audio). To tackle this challenge, we propose Omniguard, an approach for detecting harmful prompts across languages and modalities. Our approach (i) identifies internal representations of an LLM/MLLM that are aligned across languages or modalities and then (ii) uses them to build a language-agnostic or modality-agnostic classifier for detecting harmful prompts. Omniguard improves harmful prompt classification accuracy by 11.57\% over the strongest baseline in a multilingual setting, by 20.44\% for image-based prompts, and sets a new SOTA for audio-based prompts. By repurposing embeddings computed during generation, Omniguard is also very efficient ($\approx\!120 \times$ faster than the next fastest baseline). Code and data are available at: https://github.com/vsahil/OmniGuard.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。