arXiv Artificial Intelligence· Junchen Ding, Jialiang Dong, Yichen Zhu, Yi Liu, Gelei Deng, Willy Susilo, Siqi Ma, Yuekang Li·· 2 天前AI 评分37
AI Slop 与漏洞评估中的模型幻觉：推理失效与可信缓解综述
AI Slop and Hallucinations in Vulnerability Assessment: A Survey on Reasoning Failures and Trustworthy Mitigation
AI 导读
一篇被 SiMLA 2026 接收的综述指出，LLM 进入网络安全漏洞评估后，幻觉漏洞、看似合理却错误的补丁等“AI slop”正给人工分诊带来类似拒绝服务攻击的认知负担。
来源：arXiv Artificial Intelligence · arxiv.org