GitHub上64%的推送竟是僵尸网络刷量，单账号一小时狂推690次——这波暗流涌动值得细看
Every hour GH Archive records ~160K public GitHub events, and right now ~64% of the pushes are spam. Thousands of auto-created accounts (word+digits u…
用智能体方法对COTS二进制进行自动化漏洞推理，突破传统静态分析的局限，安全研究新范式。
arXiv:2605.05000v2 Announce Type: replace-cross Abstract: LLM agents have been increasingly adopted for solving security tasks. However, existing eval…
VirusTotal 通过整合70+反病毒引擎扫描链接和文件，快速识别恶意内容，用于分析文章中的可疑 URL 最合适。
Article URL: https://roman.pt/posts/linkedin-backdoor/ Comments URL: https://news.ycombinator.com/item?id=48546294 Points: 675 # Comments: 140
获CAIS最佳论文奖，研究Agent技能生态系统的安全分析，强调上下文影响
arXiv:2603.16572v2 Announce Type: replace-cross Abstract: Agent skills extend local AI agents, such as Claude Code and OpenClaw, with additional funct…
在线扫描URL并分析安全性，可快速识别伪造的验证页面和恶意脚本，帮助防范钓鱼攻击。
I just had a site pop up showing cloudflare's verify you're a human page. after checking the check box, a modal pops up telling you to 1. press comman…
系统化综述首度全面剖析GPT与DeepSeek系列模型的越狱抗性，揭示安全漏洞与防御边界
arXiv:2506.18543v2 Announce Type: replace-cross Abstract: The rapid proliferation of Large Language Models (LLMs) has heightened concerns regarding th…
DeepSource 用 AI 自动审查代码漏洞与质量问题，帮团队避免低质量危险代码的批量产生
IT之家 5 月 23 日消息，华尔街日报昨日（5 月 22 日）发布博文，报道称参与打造“龙虾”OpenClaw 的 2 名工程师示警， AI 不只加快写代码速度，也可能把低质量代码批量扩散到真实产品与服务里。 OpenClaw 内部智能体框架 Pi 创建者马里奥 · 泽克纳（Mario Zech…
Mullvad VPN的WireGuard出口IP分配机制可能存在确定性，跨服务器关联用户将威胁匿名性，隐私用户需警惕。
Analysis suggests Mullvad’s WireGuard exit IP assignment may be deterministic, allowing potential correlation of users across different servers. Could…
深度解析Sorry勒索软件的RSA+RSA+AES-GCM分层加密机制与完整攻击链，安全团队防“假期攻击”必读
手动内存取证的痛点：自动化还是坚持手动？技术讨论等你参与。
Do you automate your memory forensics work? If yes, How? If no, Why? Comments URL: https://news.ycombinator.com/item?id=48193226 Points: 2 # Comments:…
Vercel内置AI助手，自动分析应用性能与安全异常，定位原因并推荐操作，现已开启限测。
Vercel Agent is now available in Limited Beta . Agent is an AI assistant built into the Vercel dashboard that analyzes your app performance and securi…