Rogue ransomware affiliate poses as recovery firm to steal payments
A suspected ransomware affiliate is posing as a ransomware recovery service called "Ransom Busters," contacting the vict
2026-08-20 归档
今日共收录 1886 条安全情报，包含 1062 个 CVE，406 篇安全通告，以及 120 篇研究论文。
← 返回归档列表Hackers compromise 14,500 Dahua web cameras in 35-day campaign
In a large-scale campaign that researchers dubbed CameraSwarm, hackers compromised more than 14,500 Dahua IP cameras mos
US warns of AI-powered attacks on Siemens PLCs in critical infrastructure
U.S. cybersecurity agencies warn that threat actors are using AI-generated scripts to exploit Siemens S7 Series programm
US charges Iranian hackers over $3.4 billion intellectual property theft
The U.S. has charged 17 Iranians, alleged members of a hacking-for-hire company called Mabna Institute, involved in year
Password spraying attacks surge 155x as hackers exploit MFA gaps
Huntress observed a 155x increase in password spraying attacks in H1 2026, including a campaign that generated more than
Microsoft fixes known issue causing Windows Defender crashes
Microsoft has resolved a bug that caused Windows Defender to crash after a recent security update, resulting in 0xc00000
Windows 11 24H2 Home and Pro reach end of support in 2 months
Microsoft has reminded customers that systems running Home and Pro editions of Windows 11 24H2 will stop receiving updat
CISA: Medusa ransomware hit over 500 critical infrastructure orgs
The FBI said Tuesday that the Medusa ransomware gang has breached more than 500 critical infrastructure organizations in
Cloudflare Workers Spectre Attack Leaks JWT From Co-Located Worker at 12 Bits/Second
Cybersecurity researchers have disclosed details of a remote Spectre attack against Cloudflare Workers that leaked a JSO
OpenAI Pauses Frontier RL Training as It Tightens Defenses Against Unsafe AI Behavior
OpenAI on Tuesday revealed that it paused reinforcement learning (RL) training for its latest artificial intelligence (A
Clop-Linked Windchill Web Shell Decrypts Credentials and Maps Engineering Data
A JavaServer Pages (JSP) web shell deployed following the exploitation of a critical security flaw in PTC Windchill and
Cl0p Ransomware Group Names Over 40 Victims of PTC Windchill Campaign
The cybercrime gang has listed major companies such as Shell, Philips, Fiserv, Zebra, Mindray, and Largan Precision. The
CareCloud Data Breach Impact Grows to 3.7 Million Individuals
The data breach was initially believed to affect roughly 350,000 people, but the HHS breach tracker shows a far bigger i
Topological Attribution Distance (TAD): Revealing Segment-Level RAG Influence on LLM Output Geometry for Incident Log Analysis
本文提出了一种名为拓扑归因距离（Topological Attribution Distance, TAD）的新方法，用于解决大语言模型（LLM）在网络安全运营中输出证据归因与溯源的问题。随着LLM和Agentic AI被集成到安全运维系统
What to Remember, What to Reveal: Privacy-Aware Memory for Conversational Agents
本文针对基于大语言模型（LLM）的个性化对话代理中长期记忆模块的隐私风险展开研究。现有记忆架构通常只优化功能效用（如个性化响应质量），却忽视了不必要存储和重用个人隐私属性（如PII）所带来的泄露风险。仅仅删除敏感值会降低系统效用，因此隐私保
FlowShield: cryptocurrency anti-money laundering with transaction semantics parsing and fund flow tracking
加密货币反洗钱（Crypto AML）正面临日益复杂的洗钱手法，攻击者通过多种交易语义和多条区块链快速分割、转移被盗资产。现有方法往往简化交易语义、过度依赖拓扑结构特征，或仅输出孤立的检测标签，难以应对跨链、多跳的资金转移。针对这一问题，本
BullsEye: Directed Firmware Fuzzing
本文提出BULLSEYE，首个面向闭源Linux固件的定向灰盒模糊测试(DGF)框架，以基本块级距离调度模糊测试，精准定位用户指定目标。背景是IoT设备广泛使用，第三方组件复用导致已知漏洞混入固件，但确定特定镜像是否暴露漏洞十分困难，公开P
COMIC: Reference-Aware Safety Gating for Multimodal Large Language Models
该论文聚焦于多模态大语言模型（MLLMs）在视觉输入（如截图、扫描文档、图表）场景中的新型安全风险。作者指出，现有的多模态越狱攻击中，单独的提示词或图像通常无害，但当模型将看似良性的操作（如摘要、翻译、跟随指令）绑定到图像中的特定局部目标时
Benchmarking the Benchmarks: Evaluating Automated Safety Benchmarks for Small Language Models
本文针对小型语言模型（SLM）在资源受限和隐私敏感场景中日益广泛的应用，提出了一个关键问题：现有的AI安全/安全/合规基准测试主要面向大型语言模型（LLM），是否能够有效且可靠地评估SLM？作者对五个广泛使用的基准套件和26个开源SLM进行
PACE: Policy-Attested Contract Execution for Safe AI Agents in Decentralized Finance
本文提出 PACE（Policy-Attested Contract Execution），一个面向去中心化金融（DeFi）中基于大语言模型（LLM）的自治智能体的交易级授权框架。研究背景是：随着 LLM 智能体被用于执行兑换、借贷、收益管