← The feed Many Minded · The brief
SECURITY · forward · impact 3/5 · 2026-11-19

AI-Found NGINX Flaws Prompt Critical Patches for 225M Websites

AI discovered critical vulnerabilities in NGINX, powering 225 million websites globally. F5 released emergency patches for 10 flaws, including two critical remote code execution risks with CVSS 9.2 ra

Researchers using AI identified critical vulnerabilities in NGINX, the open-source web server technology powering approximately 225 million websites worldwide. F5 released emergency patches addressing 10 CVEs, including two critical remote code execution flaws (CVE-2026-123633 and CVE-2026-123007) rated at CVSS 9.2. Both vulnerabilities require HTTP/2 to be enabled on NGINX servers for exploitation, with CVE-2026-123007 affecting versions from 1.9.5 (2015) onward. Xint demonstrated attacks using these flaws to execute commands on NGINX workers and escalate to root privileges across AWS, Azure, Google Cloud, and bare-metal servers.

The vulnerabilities were discovered during preparation for ZeroDay Cloud 2026 and independently identified by multiple security researchers. F5 recommends immediate updates to NGINX 1.31.9 for affected systems. While the source states no evidence of active exploitation in the wild, the flaws represent a growing trend of AI-discovered critical flaws in foundational technologies like PHP, Apache, and Java.

This patching effort directly strengthens digital infrastructure resilience for billions of users. By securing NGINX—a core component of web services—critical disruptions to internet access and data integrity are mitigated for the Connection sector. For users relying on free web services, this reduces the risk of service outages and data breaches that could otherwise gate access to essential online resources.

What to watch: Patching rates across the 225 million websites, potential for active exploitation as systems update, and whether this AI-driven vulnerability discovery accelerates broader security improvements. Caveats: No confirmed active exploitation exists, the flaws were found via AI-assisted research, and not all affected systems have been patched.

Source: The Register