用图结构优化恶意提示生成，提升大模型安全测试效率
Graph of Attacks with Pruning: Optimizing Stealthy Jailbreak Prompt Generation for Enhanced LLM Content Moderation
- 构建攻击图谱实现多路径知识共享，减少无效尝试
- 攻击成功率超96%，查询成本降低62.7%
- 适合安全研究人员和内容审核系统开发者
随着大语言模型广泛应用，其对抗性滥用的防御能力至关重要。本文提出GAP（Graph of Attacks with Pruning）框架，用于生成隐蔽的越狱提示以评估和增强大模型防护机制。GAP通过互联图结构克服传统树形方法的局限，实现攻击路径间知识共享。实验表明，GAP相较现有技术攻击成功率提升20.8%，查询成本降低62.7%。在开放与封闭模型上均表现优异，攻击成功率超过96%。此外，我们设计了GAP-Auto（自动种子生成）与GAP-VLM（多模态攻击）等变体。使用GAP生成的提示微调内容审核系统，可使真实阳性检测率提升108.5%，准确率提高183.6%。代码已开源：https://github.com/dsbuddy/GAP-LLM-Safety。
原文摘要 · Abstract (English)
As large language models (LLMs) become increasingly prevalent, ensuring their robustness against adversarial misuse is crucial. This paper introduces the GAP (Graph of Attacks with Pruning) framework, an advanced approach for generating stealthy jailbreak prompts to evaluate and enhance LLM safeguards. GAP addresses limitations in existing tree-based LLM jailbreak methods by implementing an interconnected graph structure that enables knowledge sharing across attack paths. Our experimental evaluation demonstrates GAP's superiority over existing techniques, achieving a 20.8% increase in attack success rates while reducing query costs by 62.7%. GAP consistently outperforms state-of-the-art methods for attacking both open and closed LLMs, with attack success rates of >96%. Additionally, we present specialized variants like GAP-Auto for automated seed generation and GAP-VLM for multimodal attacks. GAP-generated prompts prove highly effective in improving content moderation systems, increasing true positive detection rates by 108.5% and accuracy by 183.6% when used for fine-tuning. Our implementation is available at https://github.com/dsbuddy/GAP-LLM-Safety.
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。