今日共收录 996 条安全情报，包含 192 个 CVE，454 篇安全通告，以及 95 篇研究论文。
← 返回归档列表Lloyds Banking Group is confident in its ability to detect sophisticated attacks and can now focus on what matters most
Google is a Leader in the Forrester Wave™: Cybersecurity Incident Response Services, Q2 2024Read the report
Google is a Leader in the IDC MarketScape: Worldwide Cybersecurity Consulting Services 2024 Vendor AssessmentRead the re
Cyber Defense Summit 2026Register today to reserve your spot
Cyber Defense Summit 2026Crafted to equip elite security professionals with the strategies, tools, and insights needed t
Security TalksJoin our security experts in this ongoing series as they explore the latest AI innovations across our secu
New Abuse of the ClickOnce Technology, Part 2: Stop Threat Actors from Clicking Once and Staying Forever
New Abuse of the ClickOnce Technology, Part 1: The Inner Workings of ClickOnce Application Deployment
Browser Security: Zero-Days Are Only Part of the Problem
A look inside the reverse-engineering journey of building the first RDP client outside of Windows to support WebAuthn re
What do board games and cybersecurity have in common? Pattern recognition. Strategy. Adaptation. In this week’s Threat S
A joint operation involving Google has disrupted NetNut, a residential proxy network that gave access to millions of com
A new phishing-as-a-service (PhaaS) platform dubbed "ARToken" appears to operate as an affiliate of the EvilTokens phish
Anthropic says Claude Fable 5 won't be accessible via Claude subscriptions after July 7, but it's not a permanent change
Claude Fable, the company's most powerful model, is now available to all users, but early impressions are disappointing,
Cybersecurity researchers have discovered a previously undocumented modular malware framework codenamed Avalon that's di
Attack demonstrates how LLM agents can combine known exploitation techniques with real-time reasoning to automate comple
大型语言模型（LLM）正在迅速改变软件开发，但其在安全关键场景中的应用引发了一个关键问题：模型是否知道其生成的代码何时不安全？这种属性被称为校准（calibration），衡量模型的置信度是否与其输出的真实正确性一致。本文首次对LLM生成代
该论文提出了一种针对函数调用型大语言模型的新型越狱攻击框架SMT（Simulated Moderation Traces）。研究者指出，在支持函数调用的LLM应用中，开发者定义的模式、结构化参数以及不可信的工具输出被交错混合到单个共享模型上
本研究针对众包平台（如Prolific和Mechanical Turk）上调查问卷参与者越来越多使用大型语言模型（LLM）辅助回答的现象，旨在量化这种行为的普遍性并探索检测与缓解方法。研究团队开展了一系列调查（样本量N=250），系统评估了