In November 2026, the information security community faced one of the most severe incidents involving infrastructure software: five independent research groups — Wiz Research, Ikotas Labs, Metnir, Rapid7, and Xint — almost simultaneously discovered two critical vulnerabilities in Nginx, one of the most widespread open-source web servers. Both bugs, designated as CVE-2026-123633 and CVE-2026-123007, allow an attacker to remotely execute arbitrary code on a target machine without any prior authentication step. On the CVSS 4.0 scale, both vulnerabilities are rated 9.2, classifying them as critical. Servers hosting websites, web applications, and APIs are at risk — effectively a significant portion of the internet infrastructure.
Attack Mechanism: From Use-After-Free to Integer Overflow
The first vulnerability, CVE-2026-123633, is a use-after-free memory management error: the server releases a memory region that is still being used during the processing of an incoming HTTP request. The second, CVE-2026-123007, stems from an integer overflow in the HTTP/2 protocol flow control mechanism, leading to incorrect memory operations. Both errors are directly related to HTTP/2 processing within the Nginx core, rather than additional modules or user code. This fundamentally distinguishes them from typical web application vulnerabilities: a successful attack does not require bugs in the code of the site or application itself. Researchers confirmed the possibility of code execution even on a server that merely serves static files without any backend components.
Eleven Years in the 'Blind Spot': The 2015 Vulnerability
The chronology of CVE-2026-123007 draws special attention: this bug has been present in the Nginx code since version 1.9.5, released in 2015. Thus, the vulnerability remained unnoticed for over a decade until discovered in 2026. This highlights the complexity of auditing mature codebases where millions of lines of code undergo minimal changes over long periods. Having gained control over the server, an attacker can steal confidential data, modify website content, or completely disrupt services, making timely updates critical.
Detection Toolkits: AI Assistants and Proprietary Platforms
Wiz Research, one of the first groups to publicly describe both vulnerabilities, used its proprietary Wiz Atlas analytical platform combined with Gemini 4 Carbon and ChatGPT 5.6 Sol generative models to verify the exploitation impact. The researchers confirmed the possibility of remote code execution and prepared proof-of-concept exploits, though they did not plan to disclose them at the time of publication. Independent confirmation from Ikotas Labs, Metnir, Rapid7, and Xint reinforces the reliability of the findings and rules out the possibility of a false positive from a single specific tool.
F5 Fixes: Which Versions Close Both Holes
F5, responsible for the development and maintenance of Nginx, promptly prepared patches. For Nginx Open Source, both vulnerabilities are resolved in versions 1.30.7 and 1.31.9. For commercial Nginx Plus, fixed builds have been released: 37.0.7.1 and 37.2.0.1 close CVE-2026-123633, while 37.0.8.1 and 37.3.0.1 address CVE-2026-123007. Infrastructure administrators are strongly advised to immediately check their installed Nginx versions and apply the corresponding updates, especially if servers accept external HTTP/2 traffic.
Practical Recommendations for Administrators
Given that the vulnerabilities affect the HTTP/2 processing core and do not depend on application configuration, standard web application-level defenses (WAF, request filtering) do not guarantee complete protection. The priority action is to update Nginx itself to the specified versions. As a temporary measure until the update can be applied, disabling HTTP/2 support on affected instances can narrow the attack surface, although it does not eliminate the vulnerability itself. Monitoring server logs for anomalous HTTP/2 request patterns will also help detect exploitation attempts before they succeed.