Key Notes
- Wiz says its Atlas system discovered two NGINX bugs, with Gemini and OpenAI models helping validate unauthenticated code execution.
- Both flaws require HTTP/2 to be enabled, so the 225 million websites using NGINX are not all necessarily exposed.
- Fixes are available in NGINX 1.30.7 stable and 1.31.9 mainline, released on November 18.
Wiz says AI helped uncover two NGINX vulnerabilities that can let unauthenticated attackers execute code on affected servers, including a flaw dating back to 2015.
NGINX released fixes on November 18 in mainline version 1.31.9 and stable version 1.30.7. The bugs affect HTTP/2 processing, making server configuration an important part of determining exposure.
The stakes are substantial: Netcraft’s September survey counted 225.4 million websites using NGINX. That measures the software’s footprint, rather than the number of vulnerable sites. A website count also cannot be treated as a count of individual servers.
Two Bugs, Different Paths to Exposure
The older vulnerability, CVE-2026-123007, involves memory corruption in HTTP/2 flow control. NGINX’s advisory lists affected versions from 1.9.5 through 1.31.8. Version 1.9.5 shipped in September 2015, leaving the issue in released code for more than 11 years.
CVE-2026-123633 is a use-after-free problem, in which software continues using memory that has already been released. It affects versions 1.19.7 through 1.31.8. The first of those releases arrived in February 2021, almost six years ago.
Both require HTTP/2 to be enabled. The use-after-free additionally involves reverse-proxy processing when connection slots run low.
Wiz rates both flaws at 9.2 under CVSS 4.0. NGINX’s own advisory list uses the severity labels major for the flow-control issue and medium for the use-after-free, while confirming both affected-version ranges and fixes.
Where AI Entered the Investigation
In its disclosure, Wiz explains that it used Atlas, the system behind its AI code-scanning product, to discover the bugs. Researchers used Google‘s Gemini 4 Carbon and OpenAI‘s ChatGPT 5.6 Sol to validate their impact and confirm unauthenticated code execution.
That distinction matters: discovery and exploitability validation are separate steps. Finding suspicious code is not the same as establishing that an attacker can turn it into a working compromise.
The findings follow the AI-assisted Java4Shell discovery in OpenJDK, another case in which researchers reported a serious weakness in widely deployed, long-established infrastructure.
What Operators Need to Check
For open-source NGINX, the published fix points are 1.30.7 on the stable branch and 1.31.9 on mainline. Operators need to check the version and HTTP/2 configuration of the instance actually handling their traffic.
NGINX’s documentation defines an HTTP/2 on/off setting for releases from 1.25.1 onward. Older configurations can enable the protocol through a parameter on the listen directive. Temporarily disabling HTTP/2 therefore requires checking how that particular installation enables it.
Wiz recommends disabling HTTP/2 temporarily if an immediate upgrade is not possible, then re-enabling it after patching.
For organizations using NGINX in front of multiple applications, remediation belongs at that shared entry point. Updating application code alone will not update the web-server software receiving the requests.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.