iOS 18.7 and iPadOS 18.7
iOS 18.7 and iPadOS 18.7
2026-05-28 归档
今日共收录 1581 条安全情报，包含 810 个 CVE，383 篇安全通告，以及 109 篇研究论文。
← 返回归档列表Android Code Search
Android Code Search
Android Devices
Android Devices
Secure an Android device
Secure an Android device
Android 16 QPR2
Android 16 QPR2
Mobile network security
Mobile network security
iOS 26.5 and iPadOS 26.5
iOS 26.5 and iPadOS 26.5
iOS 18.7.9 and iPadOS 18.7.9
iOS 18.7.9 and iPadOS 18.7.9
iOS 16.7.16 and iPadOS 16.7.16
iOS 16.7.16 and iPadOS 16.7.16
iOS 15.8.8 and iPadOS 15.8.8
iOS 15.8.8 and iPadOS 15.8.8
MFSA 2026-49Security Vulnerabilities fixed in Firefox for iOS 151.0
MFSA 2026-49Security Vulnerabilities fixed in Firefox for iOS 151.0
MFSA 2026-52Security Vulnerabilities fixed in Firefox for iOS 151.1
MFSA 2026-52Security Vulnerabilities fixed in Firefox for iOS 151.1
Beta Channel Update for ChromeOS / ChromeOS Flex
The ChromeOS Beta channel is being updated to OS version 16667.22.0 (Browser version 149.0.7827.40) for most ChromeOS de
Chrome for Android Update
Hi, everyone! We've just released Chrome 148 (148.0.7778.215) for Android. It'll become available on Google Play over
Chrome Stable for iOS Update
Hi everyone! We've just released Chrome Stable 149 (149.0.7827.45) for iOS; it'll become available on App Store in the n
Chrome Beta for iOS Update
Hi everyone! We've just released Chrome Beta 149 (149.0.7827.46) for iOS; it'll become available on App Store in the nex
Firewall for AI
Firewall for AI
Cordyceps: Covert Control Attacks on LLMs via Data Poisoning
大型语言模型（LLM）通常使用未经过滤的文本数据集进行微调，而对手可以污染这些数据集。现有的数据投毒攻击主要依赖于固定的触发短语，这些短语可以被异常检测、干净数据正则化或在线监控等防御手段中和。本文提出了一种新的数据投毒方法，通过共享知识（
Open-Weight LLM Fine-Tuning Defenses are Susceptible to Simple Attacks
该论文研究了当前针对开源权重大语言模型（LLMs）的安全微调防御措施，这些措施旨在防止模型被用于有害目的。然而，预训练的LLMs本身已编码了大量有害知识，这意味着攻击者可能无需额外微调，仅通过提示注入或越狱方式就能诱导模型产生有害输出。论文
GraphSteal: Structural Knowledge Stealing from Graph RAG via Traversal Reconstruction
本文针对Graph RAG（图检索增强生成）系统提出了一种名为GraphSteal的结构性知识窃取攻击。Graph RAG结合知识图谱与LLM，通过检索实体、关系和多跳依赖来增强生成质量，但这种结构化知识也引入了新的隐私风险。作者证明，通过