Aggregator

A vulnerability, which was classified as problematic, was found in GitLab Enterprise Edition up to 17.10.7/17.11.3/18.0.1. Affected is an unknown function of the component IP Access Restriction. The manipulation leads to insufficient granularity of access control. This vulnerability is traded as CVE-2025-5982. It is possible to launch the attack remotely. There is no exploit available. It is recommended to upgrade the affected component.

A vulnerability was found in OctoPrint up to 1.11.1. It has been declared as problematic. Affected by this vulnerability is an unknown functionality. The manipulation leads to improper neutralization of delimiters. This vulnerability is known as CVE-2025-48879. The attack can only be done within the local network. There is no exploit available. It is recommended to upgrade the affected component.

A vulnerability was found in CyberData 011209 SIP Emergency Intercom and classified as very critical. This issue affects some unknown processing of the component File Upload. The manipulation leads to path traversal: '.../...//'. The identification of this vulnerability is CVE-2025-30515. The attack may be initiated remotely. There is no exploit available. It is recommended to upgrade the affected component.

As cyberthreats evolve, so must our defenses. One of the most effective forms of modern distributed denial-of-service (DDoS) attacks is carpet bombing—a technique that spreads low-volume attack traffic across a wide range of IP addresses within a network. Unlike traditional DDoS attacks that target a single endpoint...

​当「技术竞赛」转向「生态竞赛」。

LWN 发表的一篇文章谈论了星际译王词典软件的潜在隐私问题。用户在 Debian 上安装星际译王时会默认安装插件包 stardict-plugin，其中之一是网易的有道在线词典插件，该插件还会链接另一个在线词典 dict.cn。在 X11 下星际译王默认会启用扫描功能，监视用户的文本选择通过弹出窗口提供自动翻译。在 Wayland 下该功能被破坏了，因为 Wayland 默认禁止应用程序捕获其它应用的文本。星际译王的 Debian 包维护者 Xiao Sheng Wen 认为这种做法没有问题，指出如果用户不想使用扫描功能或有道插件，可以禁用它们。但 Vincent Lefevre 认为，涉及隐私的功能永远不应该默认启用。用户可能会在文本选择时泄漏敏感信息，比如从密码管理器中复制粘贴密码。

Prove launched Unified Authentication, an authentication solution that passively and persistently recognizes customers, no matter where they appear or how often their devices or credentials change. This launch reflects a critical market shift: identity is no longer just a tool for fraud prevention or compliance – it’s a core enabler of digital commerce. Businesses must seamlessly recognize and authenticate customers across mobile apps, desktops, laptops, call centers, kiosks, and third-party platforms. But as companies move … More →

The post Prove Unified Authentication simplifies identity verification appeared first on Help Net Security.

Уязвимость в Exchange, которая способна обнулить безопасность домена.

Trend Micro announced new agentic AI technology designed to solve the traditional pain points associated with Security Information and Event Management (SIEM). When combined with Trend’s digital twin capabilities it will help to transform security operations by proactively mitigating security risks. “As the cybersecurity stack increasingly becomes AI driven, the security data layer must evolve to support data-hungry agentic capabilities, including infusing agentic AI into core SIEM functions. Trend Vision One Agentic SIEM enters the SIEM … More →

The post Trend Micro improves SIEM performance with agentic AI appeared first on Help Net Security.

实验室特设立开放课题，资助与实验室研究方向密切相关、具有重大科学意义和广泛应用前景的研究课题。

Rapid7 found that threat actors are able to purchase low-cost initial access broker services, with many packages offering a variety of options

检测业务是否受到此漏洞影响，请联系长亭应急服务团队！

Простые действия, которые превращают находку в уголовное дело с реальным фигурантом.

What kills most companies in a breach isn’t the first break-in. It’s what happens next. One stolen password turns into ten compromised systems, and suddenly your backups are toast, your finance apps are locked, and the help desk is sprinting with a garden hose toward a building fire. Attackers can pivot inside a network in […]

The post How Attackers Move Laterally, and How to Stop Them appeared first on ColorTokens.

The post How Attackers Move Laterally, and How to Stop Them appeared first on Security Boulevard.

在亚马逊和巴拿马等热带地区，即使在几乎未受破坏的雨林中，某些鸟类的种群数量也下降了多达 90%，而愈发强烈的极端高温很可能是主要因素。一项新研究发现，在 1950 年至 2020 年间，极端高温的加剧导致热带地区陆栖鸟类的丰度下降了 25%-38%。研究人员以地球生命力数据库中的全球陆栖鸟类种群数据作为研究起点，未包括水鸟或海鸟。随后他们从全球环境历史数据库获取了有关栖息地破坏的数据，并从欧洲中期天气预报中心获取了天气和气候的历史数据。比较所有数据后，他们发现，在南北纬 21°-43°之间的中纬度地区，栖息地破坏是导致鸟类种群数量下降的主要因素，这与其他研究一致。然而在热带地区，极端高温才是最大的原因。研究人员表示，在这些地区，鸟类通常生活在接近其耐热极限的边缘，一旦超过这个极限就会死亡。即使能在极端高温事件中幸存下来，变差的健康也会减少它们繁殖的机会。

Microsoft announced today that systems running Home and Pro editions of Windows 11 23H2 will stop receiving updates in three months. [...]

Rubrik launched Agent Rewind, following the close of Rubrik’s acquisition of Predibase. Agent Rewind, powered by Predibase AI infrastructure, will enable organizations to undo mistakes made by agentic AI by providing visibility into agents’ actions and enabling enterprises to rewind those changes to applications and data. “As companies consider investing in AI, they often don’t take into account the mistakes that AI agents can and will make,” said Johnny Yu, Research Manager at IDC. “Agentic … More →

The post Rubrik Agent Rewind enables organizations to undo mistakes made by agentic AI appeared first on Help Net Security.

最近几个月 AI 公司开始转向模拟推理模型，使用思维链通过多个逻辑步骤解决难题。但模拟推理真的是推理吗？已有研究显示，如果一个问题中包含上下文无关的文本，模型出错的可能性将会大增。根据发表在 arxiv 上的一篇预印本，亚利桑那大学的研究人员认为，思维链模型只是类推理文本的模拟器。他们的测试发现，思维链模型所谓的性能飞跃只是一种脆弱的幻觉，它展示的只是对训练过程中所学到的模式的复制，而不是真正的对文本的理解。思维链模型没有表现出广义的逻辑推理能力，而是展现出一种复杂的结构化模式匹配形式。稍稍偏离其训练分布，性能就会显著下降。模型生成流畅但胡扯的语言的能力创造出一种虚幻的信任光环，其内容经不起仔细审查。研究人员警告不要将思维链模型的输出等同于人类思维，不要在医学、金融或法律分析等高风险领域过于信任大模型。

A significant number of Citrix NetScaler devices continue to pose serious security risks, with approximately 7,000 systems still vulnerable to two critical exploits that have been added to the Cybersecurity and Infrastructure Security Agency’s (CISA) Known Exploited Vulnerabilities (KEV) catalog. The ongoing exposure highlights persistent challenges in enterprise patch management and cybersecurity hygiene. Widespread Network […]

The post 7,000 Citrix NetScaler Devices Still Vulnerable to CVE-2025-5777 and CVE-2025-6543 appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.