- by x32x01 ||
Two critical security vulnerabilities discovered in NGINX could allow attackers to execute arbitrary code on vulnerable servers without authentication. Disclosed on November 18, 2026, the vulnerabilities affect NGINX's HTTP/2 processing and could expose websites, APIs, and sensitive server data.
The research team at Wiz and several independent security researchers identified the flaws, which are tracked as
NGINX powers more than 225 million websites worldwide. However, this does not mean that 225 million websites are vulnerable or compromised. Actual exposure depends on the NGINX version, configuration, and whether HTTP/2 is enabled.
RCE means an attacker may be able to execute commands on a remote server without having a legitimate account or logging in.
NGINX is commonly used as:
The problem can occur during connection management, particularly when available connection slots become limited and NGINX attempts to close reusable connections.
As a result, the request-processing code may continue accessing memory that has already been freed.
This creates a
NGINX tracks the amount of response data it can send using an internal counter called
Under certain conditions, canceling a stream can cause NGINX to return unused flow-control credit without properly checking the maximum allowed value.
This can push the counter beyond its valid limit.
When NGINX subsequently performs its boundary checks, an integer underflow can cause an invalid value to be accepted. That value can eventually lead to an out-of-bounds memory copy.
Wiz researchers demonstrated that this issue could be exploited to achieve unauthenticated remote code execution, including on servers delivering ordinary static files without a backend application or third-party modules.
Depending on the server's permissions and configuration, this could allow attackers to:
According to Wiz, the fixed releases are:
These fixes address both vulnerabilities in the respective release branches.
Compare the reported version with the fixed releases and check your distribution's security updates if NGINX was installed through a Linux package manager.
For NGINX 1.25.1 and later, add the following directive inside the affected
For older configurations, remove the
Test the configuration and reload NGINX:
Make sure the configuration is compatible with your applications and that the old worker processes have exited.
Disabling HTTP/2 is a temporary mitigation for these two vulnerabilities, not a replacement for installing the security updates.
The research also involved independent discoveries by security teams at Ikotas Labs, Metnir, Rapid7, and Xint.
An interesting aspect of this research is the role of artificial intelligence in vulnerability discovery.
Wiz used its Wiz Atlas research environment, alongside AI models, to help identify and validate complex memory-safety problems that had remained in NGINX for years.
One of the vulnerabilities, CVE-2026-123007, traces back to NGINX 1.9.5, released in 2015. This means the underlying flaw had existed for more than a decade before its public disclosure.
This highlights how AI-assisted security research can uncover weaknesses in widely deployed software, even when that software has undergone years of development and review.
Although the 225-million figure should not be interpreted as the number of compromised websites, the potential impact is significant.
If you manage an NGINX server, check its version, review its HTTP/2 configuration, and prioritize installing the fixed releases. Do not rely on disabling HTTP/2 as a permanent security solution.
Security References:
The research team at Wiz and several independent security researchers identified the flaws, which are tracked as
CVE-2026-123633 and CVE-2026-123007.NGINX powers more than 225 million websites worldwide. However, this does not mean that 225 million websites are vulnerable or compromised. Actual exposure depends on the NGINX version, configuration, and whether HTTP/2 is enabled.
🔍 What Is NGINX2Shell?
NGINX2Shell is the name associated with two critical vulnerabilities in the NGINX web server that can lead toRemote Code Execution (RCE).RCE means an attacker may be able to execute commands on a remote server without having a legitimate account or logging in.
NGINX is commonly used as:
- 🌐 A web server that delivers website content.
- 🔄 A reverse proxy that forwards requests to application servers.
- ⚖️ A load balancer that distributes traffic across multiple servers.
- 🔐 An HTTPS endpoint that handles encrypted connections.
🚨 The Two Critical NGINX Vulnerabilities
Both vulnerabilities involve memory-safety problems in NGINX's HTTP/2 implementation. They have a reported CVSS 4.0 score of 9.2 out of 10, placing them in the Critical severity category.| CVE | Vulnerability type | Potential impact |
|---|---|---|
| CVE-2026-123633 | Use-After-Free (UAF) | Unauthenticated remote code execution |
| CVE-2026-123007 | Integer Underflow | Out-of-bounds memory operations and remote code execution |
1. CVE-2026-123633: HTTP/2 Use-After-Free
This vulnerability occurs when NGINX releases memory associated with an HTTP/2 connection while an active request still references that memory.The problem can occur during connection management, particularly when available connection slots become limited and NGINX attempts to close reusable connections.
As a result, the request-processing code may continue accessing memory that has already been freed.
This creates a
Use-After-Free (UAF) condition that attackers can potentially exploit to execute arbitrary code on the server.2. CVE-2026-123007: HTTP/2 Integer Underflow
The second vulnerability involves incorrect HTTP/2 flow-control accounting.NGINX tracks the amount of response data it can send using an internal counter called
send_window.Under certain conditions, canceling a stream can cause NGINX to return unused flow-control credit without properly checking the maximum allowed value.
This can push the counter beyond its valid limit.
When NGINX subsequently performs its boundary checks, an integer underflow can cause an invalid value to be accepted. That value can eventually lead to an out-of-bounds memory copy.
Wiz researchers demonstrated that this issue could be exploited to achieve unauthenticated remote code execution, including on servers delivering ordinary static files without a backend application or third-party modules.
💥 What Can Attackers Do?
Successful exploitation could give an attacker control over the NGINX worker process running on an affected server.Depending on the server's permissions and configuration, this could allow attackers to:
- Execute unauthorized commands on the server.
- Access sensitive information and application data.
- Modify website files or content.
- Compromise applications and APIs served through NGINX.
- Disrupt website availability and server operations.
🛡️ How to Protect Your NGINX Server
Administrators should identify affected NGINX installations and apply the security updates released by F5.According to Wiz, the fixed releases are:
| Product | Fixed versions |
|---|---|
| NGINX Open Source (Stable) | 1.30.7 |
| NGINX Open Source (Mainline) | 1.31.9 |
| NGINX Plus | 37.0.7.1 and 37.2.0.1 for CVE-2026-123633; 37.0.8.1 and 37.3.0.1 for CVE-2026-123007 |
1. Check Your NGINX Version
First, identify the installed NGINX version by running: Bash:
nginx -v 2. Temporarily Disable HTTP/2
If you cannot install the security updates immediately, Wiz recommends temporarily disabling HTTP/2 on affected NGINX instances.For NGINX 1.25.1 and later, add the following directive inside the affected
server block: Code:
http2 off; http2 parameter from the relevant listen directives.Test the configuration and reload NGINX:
Bash:
sudo nginx -t
sudo systemctl reload nginx Disabling HTTP/2 is a temporary mitigation for these two vulnerabilities, not a replacement for installing the security updates.
3. Monitor Suspicious Activity
Administrators should investigate unusual behavior, including:- Unexpected NGINX worker crashes or restarts.
- Unusual commands launched by NGINX worker processes.
- Suspicious activity involving long-lived HTTP/2 connections.
- Unexpected changes to website files or server configurations.
🧠 How Were These Vulnerabilities Discovered?
Wiz Research worked with F5, the company behind NGINX, to investigate and responsibly disclose the vulnerabilities.The research also involved independent discoveries by security teams at Ikotas Labs, Metnir, Rapid7, and Xint.
An interesting aspect of this research is the role of artificial intelligence in vulnerability discovery.
Wiz used its Wiz Atlas research environment, alongside AI models, to help identify and validate complex memory-safety problems that had remained in NGINX for years.
One of the vulnerabilities, CVE-2026-123007, traces back to NGINX 1.9.5, released in 2015. This means the underlying flaw had existed for more than a decade before its public disclosure.
This highlights how AI-assisted security research can uncover weaknesses in widely deployed software, even when that software has undergone years of development and review.
❓ Frequently Asked Questions
-----------------------Are 225 million websites affected by NGINX2Shell?
No. NGINX powers more than 225 million websites, but this figure represents its worldwide usage, not the number of vulnerable websites. Actual exposure depends on the installed version and server configuration.Does NGINX2Shell require authentication?
No. Both vulnerabilities can potentially allow unauthenticated remote code execution when their respective exploitation conditions are met.Does disabling HTTP/2 protect against both vulnerabilities?
Disabling HTTP/2 is the temporary mitigation recommended by Wiz for these two vulnerabilities. However, upgrading to the fixed releases is the proper long-term solution.Are websites using WordPress or other CMS platforms affected?
They may be exposed if their hosting infrastructure uses a vulnerable NGINX installation. The vulnerability is in the web server, not specifically in WordPress or another CMS.Is there a publicly available exploit?
Wiz confirmed exploitation in controlled testing environments but stated that it would not release its proof-of-concept exploits at the time of disclosure.📌 Final Thoughts
NGINX2Shell highlights the risks of vulnerabilities in infrastructure software that sits at the front of millions of websites and applications.Although the 225-million figure should not be interpreted as the number of compromised websites, the potential impact is significant.
If you manage an NGINX server, check its version, review its HTTP/2 configuration, and prioritize installing the fixed releases. Do not rely on disabling HTTP/2 as a permanent security solution.
Security References: