统一威胁检测与缓解框架（UTDMF）：对抗企业级Transformer中的提示注入、欺骗与偏见
Unified Threat Detection and Mitigation Framework (UTDMF): Combating Prompt Injection, Deception, and Bias in Enterprise-Scale Transformers
AI总结：
针对企业级LLM面临的提示注入、欺骗与偏见威胁，研究扩展对抗激活补丁框架提出UTDMF，经700余次实验验证，在检测、减骗、公平性上均有显著效果，还提出新算法、假设及部署工具包。
摘要：
大型语言模型（LLM）在企业系统中的快速应用暴露出提示注入攻击、策略性欺骗和有偏输出等漏洞，威胁着安全性、可信度与公平性。我们扩展了此前的对抗激活补丁框架（arXiv:2507.09406，该框架在小型网络中诱导欺骗的成功率为23.9%），提出了统一威胁检测与缓解框架（UTDMF）——这是一个可扩展的实时流水线，适用于Llama-3.1（405B）、GPT-4o、Claude-3.5等企业级模型。通过对每个模型开展700余次实验，UTDMF实现了：（1）对提示注入（如越狱攻击）92%的检测准确率；（2）通过增强补丁将欺骗输出降低65%；（3）公平性指标（如人口统计偏见）提升78%。其创新贡献包括用于多威胁检测的通用补丁算法、三个关于威胁交互（如企业工作流中的威胁链）的突破性假设，以及一套带有API、可直接部署用于企业集成的工具包。
英文摘要：
The rapid adoption of large language models (LLMs) in enterprise systems exposes vulnerabilities to prompt injection attacks, strategic deception, and biased outputs, threatening security, trust, and fairness. Extending our adversarial activation patching framework (arXiv:2507.09406), which induced deception in toy networks at a 23.9% rate, we introduce the Unified Threat Detection and Mitigation Framework (UTDMF), a scalable, real-time pipeline for enterprise-grade models like Llama-3.1 (405B), GPT-4o, and Claude-3.5. Through 700+ experiments per model, UTDMF achieves: (1) 92% detection accuracy for prompt injection (e.g., jailbreaking); (2) 65% reduction in deceptive outputs via enhanced patching; and (3) 78% improvement in fairness metrics (e.g., demographic bias). Novel contributions include a generalized patching algorithm for multi-threat detection, three groundbreaking hypotheses on threat interactions (e.g., threat chaining in enterprise workflows), and a deployment-ready toolkit with APIs for enterprise integration.
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。
