用「稀有区域指纹」精准追踪大模型血统，一篇学术新方法直接看。
arXiv:2505.12682v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly released under restricted licenses, creating a growi…
让AI Agent自动互相攻击，发现生产环境漏洞，这篇论文提出了全新的自动化红队测试框架。
arXiv:2607.11698v1 Announce Type: cross Abstract: Production LLM agents such as Claude Code and Codex operate over untrusted content, files, commands,…
提出选择性安全引导方法，通过价值过滤解码提升LLM安全性
arXiv:2605.14746v2 Announce Type: replace Abstract: While large language models (LLMs) are trained to align with human values, their generations may s…
揭秘大模型安全探针为何在最终token失效：早于最后一层的内部表征已暴露漏洞
arXiv:2605.12726v2 Announce Type: replace Abstract: Final-token safety probes monitor a single hidden state after prompt prefill, but jailbreak prompt…
ICML 2026接收：为LLM水印引入功率校准，告别传统启发式调优。
arXiv:2607.05694v1 Announce Type: cross Abstract: Logit-based watermarking is a widely used mechanism for identifying LLM generated content, yet its e…
通过截断思维链审计，揭示LLM教育导师是否存在"答案驱动"的推理偏差，提升AI教学可信度。
arXiv:2607.04572v1 Announce Type: new Abstract: Large language model (LLM) tutors often produce fluent step-by-step explanations, but a correct and pe…
论文提出在线安全监控LLM的新方法，引入假设检验框架，为模型部署提供实时风险检测思路。
arXiv:2607.02510v1 Announce Type: new Abstract: Despite alignment training, LLMs remain prone to generating unsafe outputs at deployment time. Monitor…
最新研究发现，通过模拟审核轨迹可绕过函数调用LLM的安全限制，揭示大模型越狱攻击新路径。
arXiv:2607.00481v1 Announce Type: cross Abstract: Jailbreak attacks remain a critical threat to the safe deployment of large language models (LLMs). W…
发现LLM服务新漏洞：利用特殊Token操纵可越狱在线大模型，揭秘攻击原理与防范思路。
arXiv:2510.10271v2 Announce Type: replace-cross Abstract: Unlike regular tokens derived from existing text corpora, special tokens are artificially cr…
利用中间层熵的动态变化检测大模型越狱，为AI安全提供新思路。
arXiv:2606.25182v1 Announce Type: cross Abstract: Jailbreak attacks reveal a persistent weakness in aligned Large Language Models: carefully crafted p…
LLM智能体隐私保护新方案：上下文对齐机制，让AI更懂隐私边界。
arXiv:2606.21710v1 Announce Type: cross Abstract: AI agents acting on behalf of users are constantly making decisions, and for users to trust their ag…
提出基于不确定性的大模型去偏与遗忘新方法，有效解决数据污染问题。
arXiv:2606.23313v1 Announce Type: cross Abstract: Benchmark-based evaluation is the dominant paradigm for assessing large language model (LLM) capabil…
OpenAI新部署模拟器，提前捕获大模型不良行为，让AI安全更可靠
Article URL: https://openai.com/index/deployment-simulation/ Comments URL: https://news.ycombinator.com/item?id=48581468 Points: 1 # Comments: 0
从攻击成功率到攻击成功概率，系统性评估开源大模型提示攻击漏洞，为安全防护提供新视角。
arXiv:2505.14368v2 Announce Type: replace-cross Abstract: Recent studies demonstrate that Large Language Models (LLMs) are vulnerable to attacks that …
自动化攻击技术直击大模型多道防线，AI安全攻防再升级。
arXiv:2606.16751v1 Announce Type: cross Abstract: Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks.…
大模型安全新方案：通过神经元选择性调优NeST精准提升LLM安全性，不牺牲性能。
arXiv:2602.16835v2 Announce Type: replace-cross Abstract: Safety alignment is essential for the responsible deployment of Large Language Models (LLMs)…
论文提出PI-Hunter，自动发现并精准定位大模型提示注入漏洞，为AI安全红队测试提供新方案。
arXiv:2606.12737v1 Announce Type: cross Abstract: Large Language Models (LLMs) are rapidly evolving into agentic systems that interact with external t…
利用强化学习自动化实现提示注入攻击，性能超越人工红队测试
arXiv:2602.05746v2 Announce Type: replace-cross Abstract: Prompt injection is a critical vulnerability in LLM agents, yet the strongest methods still …
用假后门骗过真后门：无需已知信息，通过共享内部机制清除生成式大模型里的未知后门。
arXiv:2606.11648v1 Announce Type: cross Abstract: Backdoor attacks pose a serious threat to the safety and reliability of Large Language Models (LLMs)…
语义基础+固定惩罚约束优化，让大模型对齐过程获得可认证的安全保障
arXiv:2510.03520v2 Announce Type: replace-cross Abstract: Ensuring safety is a foundational requirement for large language models (LLMs). Achieving an…