通过允许模型失败来防止大模型生成有害内容，效果远超现有方法。
SafeRedirect: Defeating Internal Safety Collapse via Task-Completion Redirection in Frontier LLMs
- 让模型明确获准失败，用固定输出终止任务，避免强行完成。
- 将有害内容生成率从71.2%降至8.0%，显著优于最强基线的55.0%。
- 适用于需要完成高风险任务的大模型，特别适合安全防御研究者。
内部安全崩溃（ISC）是前沿大模型在执行需包含有害内容才能正确完成的专业任务时，自发生成有害内容的失效模式，安全失败率超过95%。现有输入级防御对ISC的失败率为100%，标准系统提示防御仅部分缓解。本文提出SafeRedirect，一种系统级覆盖机制，通过重定向模型的任务完成驱动力而非抑制它来对抗ISC。SafeRedirect明确授予失败许可，规定确定性硬停止输出，并指示模型保留有害占位符未解决。在七种前沿大模型上，针对三种人工智能/机器学习相关的ISC任务类型，在单轮设置下评估，SafeRedirect将平均不安全生成率从71.2%降低至8.0%，优于最强基线的55.0%。多模型消融实验表明，失败许可与条件特异性普遍关键，其他组件作用因模型而异。跨攻击测试验证其对ISC的最先进防御能力，且在其他攻击族上的泛化性能不低于基线。代码已公开于https://github.com/fzjcdt/SafeRedirect。
原文摘要 · Abstract (English)
Internal Safety Collapse (ISC) is a failure mode in which frontier LLMs, when executing legitimate professional tasks whose correct completion structurally requires harmful content, spontaneously generate that content with safety failure rates exceeding 95%. Existing input-level defenses achieve a 100% failure rate against ISC, and standard system prompt defenses provide only partial mitigation. We propose SafeRedirect, a system-level override that defeats ISC by redirecting the model's task-completion drive rather than suppressing it. SafeRedirect grants explicit permission to fail the task, prescribes a deterministic hard-stop output, and instructs the model to preserve harmful placeholders unresolved. Evaluated on seven frontier LLMs across three AI/ML-related ISC task types in the single-turn setting, SafeRedirect reduces average unsafe generation rates from 71.2% to 8.0%, compared to 55.0% for the strongest viable baseline. Multi-model ablation reveals that failure permission and condition specificity are universally critical, while the importance of other components varies across models. Cross-attack evaluation confirms state-of-the-art defense against ISC with generalization performance at least on par with the baseline on other attack families. Code is available at https://github.com/fzjcdt/SafeRedirect.
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。