<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Red-Team on Tech News Feed</title>
    <link>https://news.dhphong.com/tags/red-team/</link>
    <description>Recent content in Red-Team on Tech News Feed</description>
    <generator>Hugo -- 0.131.0</generator>
    <language>vi</language>
    <lastBuildDate>Tue, 21 Apr 2026 00:03:07 +0700</lastBuildDate>
    <atom:link href="https://news.dhphong.com/tags/red-team/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>[NVIDIA Developer Blog] Mitigating Indirect AGENTS.md Injection Attacks in Agentic Environments</title>
      <link>https://news.dhphong.com/posts/2026-04-21-mitigating-indirect-agentsmd-injection-attacks/</link>
      <pubDate>Tue, 21 Apr 2026 00:03:07 +0700</pubDate>
      <guid>https://news.dhphong.com/posts/2026-04-21-mitigating-indirect-agentsmd-injection-attacks/</guid>
      <description>Nguồn: NVIDIA Developer Blog
Tóm tắt Khi các AI coding agent như OpenAI Codex được tích hợp vào workflow phát triển phần mềm, chúng tạo ra một attack surface mới: các file instruction như AGENTS.md có thể bị khai thác qua indirect injection. NVIDIA AI Red Team phát hiện một lỗ hổng trong Codex cho phép attacker nhúng instruction độc hại vào file AGENTS.md thông qua các malicious dependency trong supply chain.</description>
    </item>
  </channel>
</rss>
