Posts of last few hours
Please support the site operations by clicking ads.
Attackers Used Cairn With DeepSeek and GLM Models to Compromise 10 Organizations
A Chinese-speaking threat actor used the open-source Cairn framework with DeepSeek and GLM models to attack at least 19 organizations, compromising 10. Unit 42 said targets included U.S. semiconductor, defense and human rights groups, showing how accessible AI tools can speed intrusions and data theft.
A Chinese-speaking threat actor used the open-source Cairn framework with DeepSeek and GLM models to attack at least 19 organizations, compromising 10. Unit 42 said targets included U.S. semiconductor, defense and human rights groups, showing how accessible AI tools can speed intrusions and data theft.
https://ransomware.databreachtoday.com/chinese-actor-automates-attacks-on-us-defense-chip-firms-a-33392
AuthMind Discovers Sanctioned, Unsanctioned and External AI Agents at Network Layer
IBM acquired identity security startup AuthMind to discover sanctioned and rogue AI agents, continuously monitor their behavior and revoke access when activity exceeds authorized boundaries, extending IBM Verify and HashiCorp Vault controls across the agent identity life cycle.
IBM acquired identity security startup AuthMind to discover sanctioned and rogue AI agents, continuously monitor their behavior and revoke access when activity exceeds authorized boundaries, extending IBM Verify and HashiCorp Vault controls across the agent identity life cycle.
https://ransomware.databreachtoday.com/ibm-purchases-authmind-to-detect-agents-operating-off-scope-a-33393
Officials Say Cyberattack Is Unlikely as Outage Disrupts More Than 2,000 Flights
A nationwide FAA telecommunications outage grounded more than 2,000 US flights and disrupted air traffic control centers. It came days after the agency deployed its AI SMART platform, but the cause of the outage is still unclear. Officials say a cyberattack is unlikely.
A nationwide FAA telecommunications outage grounded more than 2,000 US flights and disrupted air traffic control centers. It came days after the agency deployed its AI SMART platform, but the cause of the outage is still unclear. Officials say a cyberattack is unlikely.
https://ransomware.databreachtoday.com/us-flights-halted-as-faa-communications-systems-fail-a-33394
香港迪士尼乐园约 1.7 万张电子门票在票务系统中被盗,园方已报警并暂停网上售票及酒店预订服务。官网声明,旅客需要购票,欢迎到合作旅行社及平台或正门入口售票处。报道称,乐园门票由 Cityline 负责网上票务售卖,园方资讯科技部于 11 月 12 日晚侦测到公司电脑系统有异常活动,发现一万七千个二维码在网上票务系统被盗。香港资讯科技商会荣誉会长方保侨称:以往黑客是写程序或手动追漏洞,现在 AI 发展令黑客用 AI 寻找网络漏洞,从而攻击,近期这么多事件,我们相信只会愈来愈多。因为 AI 攻击太快,人手根本抵挡不到,开始提倡用 AI 防守,而市面上亦有这些方案。
https://www.solidot.org/story?sid=85891
После первых галактик за дело взялись прожорливые гиганты.
https://www.securitylab.ru/news/580149.php
https://mp.weixin.qq.com/s?__biz=MTMwNDMwODQ0MQ==&mid=2653116488&idx=2&sn=18938311eab6f6d73e1dbbab000047a0
冰箱、手表、落地灯,都被加上了 Al。
https://mp.weixin.qq.com/s?__biz=MTMwNDMwODQ0MQ==&mid=2653116488&idx=1&sn=74296fd963161f7dad05334dd1c2e1c4
日本环球晶圆被黑客攻击,内部资料直接在网上泄露了。
这家是全球第三大硅片厂。上周三部分工厂系统遭黑客攻击,一度中断。公司说今天多数工厂能恢复发货,但麻烦没完。
泄露包足足 117.7MB,Telegram 上已经有人传开了。里面全是日本分部的底牌:中期计划、生产数据、客户状况、库存表全在;还有质量手册、系统设计书和公司介绍。
最扎眼的是一份报价单。2022 到 2024 年累计金额接近 6.91 亿元,自费人次 184 万出头。另一份系统设计书连权限审批流程都写得清清楚楚——谁代签、谁能代理承认可是一目了然。
文章最后还留了个活口:每多 5 个转发就再放一份数据共享,地址私信要,关键词“2026.11.15”。说白了,这就是在用泄露的数据裂变引流。
https://mp.weixin.qq.com/s?__biz=MzU4NjY0NTExNA==&mid=2247490147&idx=1&sn=4fb83eba68ac540c5b86f11bc43d63ff
对斯里兰卡野生亚洲象的一项为期 12 年的研究显示,种群数量的逐渐减少可令动物社群跨越一个社会临界点,从而使长期形成的关系破裂,群体变得更小且更孤立,并威胁物种的长期生存。研究人员分析了斯里兰卡 443 头野生雌性亚洲象为期 12 年的个体观察数据。M他们发现,种群个体的逐渐流失使这一亚洲象种群跨越了社会临界点,从根本上重塑了其社会网络。随着死亡和永久离群的个体日益超过新加入的个体,象群规模变小、群体数量增多且日益孤立;到2018 年时,在最初被追踪的雌象中,近 69% 已从研究种群中消失。研究结果显示,这种碎片化不仅由个体丧失驱动,也由留存大象之间较弱社会纽带的断裂所驱动,从而导致整个种群个体间的联系减少。
https://www.solidot.org/story?sid=85890
Нейросеть позволяет свободно гулять по созданным пространствам и менять обстановку по своему вкусу.
https://www.securitylab.ru/news/580111.php
1. 漏洞概述 受影响组件为 craftcms/cms 。受影响范围取自 GitHub Security Advisory GHSA-v4q3-2g5g-wjj6 的标注:Craft 4 线为 4.0.0 至 4.18.8,Craft 5 线为 5.0.0 至 5.11.1。修复版本为 4.18.9 与 5.11.2, 5.11.2 发布说明 的 CHANGELOG 把这条按 high-severity information disclosure 归类。漏洞属于模板引擎注入(CWE-94 / CWE-1336)。触发需要两个前提:账号能登录控制面板并打开元素索引(本次实测用 admin ),且目标为 Craft 5.11.2 / 4.18.9 之前的未打补丁版本;走到 RCE 还要再加一条,该账号具备至少一个会读取 redirect 的控制面板动作权限。CVE 记录为 CVE-2026-
https://xz.aliyun.com/news/93336
Американские спецслужбы подозревают разведывательную операцию, а совет директоров отправил основателя компании в отставку после 33 лет руководства.
https://www.securitylab.ru/news/580174.php
Meta was fined €805 million in Italy yesterday over a Facebook for Android feature that suggests photos and videos from users' phone galleries for them to post.
The Garante, Italy's data protection authority, found that the feature was enabled by default and that the app accessed Italian users' photos, videos, and related metadata without their explicit consent.
The company has 60 days to
https://thehackernews.com/2026/11/italy-fines-meta-805-million-over.html
Самое неожиданное началось уже после того, как пропажу обнаружили.
https://www.securitylab.ru/news/580136.php
OpenAI发布GPT-6 Cyber模型专为漏洞挖掘设计,成本降至千美元级别,在Linux内核中发现948个本地提权漏洞
https://mp.weixin.qq.com/s?__biz=MzIzMzE4NDU1OQ==&mid=2652079884&idx=3&sn=0a6b6bcc5e3ef8b7a0a27d0d4b267a74
OpenAI发布GPT-6 Cyber模型专为漏洞挖掘设计,成本降至千美元级别,在Linux内核中发现948个本地提权漏洞
https://mp.weixin.qq.com/s?__biz=MzIzMzE4NDU1OQ==&mid=2652079884&idx=2&sn=92a9b89164638b06321978005d689a06
OpenAI发布GPT-6 Cyber模型专为漏洞挖掘设计,成本降至千美元级别,在Linux内核中发现948个本地提权漏洞
https://mp.weixin.qq.com/s?__biz=MzIzMzE4NDU1OQ==&mid=2652079884&idx=1&sn=4955203b3c4b7745ccacfa6a06f44631
На кону $74 млрд.
https://www.securitylab.ru/news/580090.php
https://mp.weixin.qq.com/s?__biz=MzU4NjY0NTExNA==&mid=2247490140&idx=1&sn=a343b7e88ee97a87a65b0f547dc1ba6f
Началось всё с самого безобидного текста, который только можно отправить на сайт.
https://www.securitylab.ru/news/580132.php
Latest Blog Posts
- 13 hours 34 minutes ago
- 13 hours ago
- 13 hours ago
- 13 hours ago
- 13 hours ago
- 13 hours ago
- 13 hours 41 minutes ago
- 13 hours 42 minutes ago
- 13 hours ago
- 13 hours ago