看似严谨的LLM网络安全决策，实则经不起对抗性考验，这份研究揭示AI安全的关键盲区。
arXiv:2608.20966v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used in cybersecurity workflows, yet it remains unclea…
AI审稿系统藏漏洞，18篇论文暗嵌白字指令操控评审结果
arXiv:2507.06185v2 Announce Type: replace-cross Abstract: In July 2025, 18 academic manuscripts on arXiv contained hidden instructions that manipulate…
让LLM智能体在动态威胁中自我进化防御，告别手工安全规则，安全研究新范式。
arXiv:2608.12977v1 Announce Type: cross Abstract: The expanding operational capabilities of large language model (LLM) agents introduce sophisticated …
分片机制如何阻断大模型监督失效与对抗攻击？这篇论文给出了关键防线设计。
arXiv:2608.06422v1 Announce Type: new Abstract: Giving an LLM judge more compute does not necessarily make it check more requirements. When one call m…
音频大模型也难逃越狱攻击？GRM通过梯度比掩蔽实现效用感知攻击，兼顾攻击效果与音频质量，为多模态AI安全防御敲响警钟。
arXiv:2604.09222v2 Announce Type: replace-cross Abstract: Audio Large Language Models (ALLMs) enable spoken interaction but introduce new jailbreak vu…
针对LLM智能体长期记忆的新型投毒攻击，零散拼接恶意记录，隐蔽操控智能体行为。
arXiv:2608.01637v1 Announce Type: new Abstract: Long-term memory enables LLM agents to retain useful information across sessions, but also creates an …
一场来自DeepSeek的300,000次攻击，反而被作者改造成防御利器，大模型安全攻防的硬核实战。
Article URL: https://jesta.ai/blog/darkreasoning Comments URL: https://news.ycombinator.com/item?id=49158479 Points: 16 # Comments: 6
多模态智能体正面临“无声”入侵：环境噪音竟可伪装成恶意指令劫持AI行为，音频注入攻击拉开安全新战线
Article URL: https://arxiv.org/abs/2607.28165 Comments URL: https://news.ycombinator.com/item?id=49123267 Points: 2 # Comments: 0
针对大模型投毒攻击的检测新方法ToxScreen，通过分析模型行为判断是否被恶意篡改
arXiv:2607.26849v1 Announce Type: cross Abstract: As large language models (LLMs) are deployed in high-stakes domains, adversaries may poison training…
多轮对话破解LLM防线，分解信用分配实现高效越狱攻击，揭示AI安全新漏洞
arXiv:2607.11070v1 Announce Type: new Abstract: Modern large language models (LLMs) operate in interactive multi-turn settings, making multi-turn jail…
新研究揭示非文本目标攻击方法，探索无需文本线索的对抗样本生成，为AI安全防御提供新视角。
arXiv:2510.02999v5 Announce Type: replace-cross Abstract: Existing gradient-based jailbreak attacks on Large Language Models (LLMs) typically optimize…
攻击AI容易，判断攻击是否造成真实危害却很难。一文讲透AI安全评测的困境与出路。
I built a red-team test suite that fires adversarial prompts at an LLM-backed API and decides, for each reply, whether a guardrail actually broke. It …
揭露低资源语言与混合代码切换可绕过LLM安全防护，揭示模型安全泛化的关键漏洞。
arXiv:2607.01859v1 Announce Type: new Abstract: Safety training for large language models (LLMs) is conducted predominantly in English, leaving uncert…
一键破解语音模型？新研究系统优化多模态越狱攻击，直指spoken language model安全漏洞。
arXiv:2603.19127v2 Announce Type: replace Abstract: As Spoken Language Models (SLMs) integrate speech and text modalities, they inherit the safety vul…
AI也会“想当然”？这项研究揭示演绎推理中的刻板印象，并提出Fair-GCG缓解方法。
arXiv:2606.30989v1 Announce Type: cross Abstract: Warning: This paper contains several toxic and offensive statements. While reasoning generally impro…
新论文揭示LLM排序漏洞：两阶段token优化可操控排名，可靠性存疑
arXiv:2510.06732v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly used as rerankers in information retrieval, yet thei…
从理论层面剖析持续学习如何抵御数据投毒攻击，为鲁棒AI提供新视角。
arXiv:2606.29841v1 Announce Type: new Abstract: Continual learning (CL), where a model is trained on a sequence of data tasks, is increasingly being a…
专为内容与AI安全设计的对抗性感知大模型，直面越狱攻击与有害内容生成难题。
arXiv:2606.27632v1 Announce Type: new Abstract: As large language models are increasingly deployed in real-world systems, safety failures can still le…
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…
提示压缩已成攻击面：黑盒手段可操纵压缩后文本，让LLM智能体偏离原意，安全研究必读。
arXiv:2510.22963v4 Announce Type: replace-cross Abstract: Prompt compression is increasingly deployed in LLM agents to reduce latency and cost, but it…