Secure Bulletin Navigating the cyber sea with knowledge
Home > Articolo > Two Critical NGINX Flaws Put More Than 225 Million Sites at Risk of Pre-Auth Code Execution
Two Critical NGINX Flaws Put More Than 225 Million Sites at Risk of Pre-Auth Code Execution
Read Time:4 Minute, 1 Second

Two newly disclosed vulnerabilities in NGINX, one of the internet’s most widely deployed web servers, could allow attackers to execute code on affected systems without ever logging in. The flaws, tracked as CVE-2026-123633 and CVE-2026-123007, were disclosed on November 18, 2026, and both carry a critical severity score of 9.2 under the CVSS 4.0 scale.

Two Flaws, Two Very Different Attack Surfaces

Both vulnerabilities require NGINX to be accepting HTTP/2 connections from clients, but the specific conditions that trigger each one differ meaningfully. According to F5’s advisory, CVE-2026-123633 is a use-after-free bug — meaning a worker process can end up reusing memory after it has already been released — that affects reverse-proxy configurations where NGINX forwards HTTP/2 client requests to a backend application, and only once the worker process is running low on available connections. Depending on what that worker process can access, successful exploitation can crash it outright or, under the right conditions, allow arbitrary code execution.

CVE-2026-123007 is the more alarming of the pair in terms of reach. Wiz Research, which helped identify the issue alongside independent researchers at Ikotas Labs, Metnir, and Xint, says it demonstrated working code execution against an NGINX server that was doing nothing more exotic than serving static files — no backend application, no reverse proxy, no third-party modules involved. That means organizations reviewing their exposure can’t limit the check to API gateways and proxies; a bare-bones NGINX instance serving a static site needs the same scrutiny. The flaw affects NGINX Open Source releases going all the way back to version 1.9.5, first released in 2015.

A Widely Used Piece of Internet Plumbing

NGINX’s footprint makes both flaws significant well beyond any single website. Netcraft’s September 2026 web server survey counted more than 225 million sites running the software, which frequently handles HTTPS termination and routes incoming traffic to multiple backend applications. That architecture means a single vulnerable NGINX deployment can put several unrelated services at risk at once, even when the applications behind it are otherwise secure — a successful attack could expose sensitive data, tamper with site content, or knock a service offline, depending on what the compromised NGINX process has access to.

Researchers were careful to note what their findings do and don’t establish: while both disclosures demonstrate that remote code execution is technically achievable under the reported conditions, neither confirms that attackers have exploited either flaw against production systems in the wild. Organizations should nonetheless treat the disclosures as urgent, given how straightforward the static-file exploitation path appears to be.

Not the Same Bugs Patched in October

Administrators who already addressed a separate batch of NGINX HTTP/2 vulnerabilities that F5 disclosed in October should not assume they are covered. F5’s advisory explicitly lists NGINX Open Source 1.31.8 — a version released after October’s fixes — as still affected by CVE-2026-123633, meaning running a newer build does not automatically mean the new vulnerabilities are patched.

Which Versions Are Fixed

Fixes for both CVEs are available in NGINX Open Source 1.31.9 on the mainline branch and 1.30.7 on the stable branch. NGINX Plus customers face a more fragmented picture:

  • CVE-2026-123633 affects NGINX Open Source 1.19.7–1.30.6 and 1.31.0–1.31.8, and NGINX Plus 37.0.0–37.0.6.1 and 37.1.0–37.2.0.
  • CVE-2026-123007 affects NGINX Open Source 1.9.5–1.30.6 and 1.31.0–1.31.8, and NGINX Plus 37.0.0–37.0.7.1 and 37.1.0–37.2.0.1.
  • NGINX Plus R-series releases R34 through R36 have no fix listed at all for either CVE; F5 recommends moving to a 37.x build that covers both.

What Administrators Should Do

F5 and Wiz recommend starting with internet-facing NGINX instances before moving to internal proxies and static sites, then checking any Linux distribution packages or bundled products that ship their own version of NGINX against the supplier’s own security notices. Crucially, simply reloading NGINX’s configuration is not enough — administrators need to confirm the actual running binary has been replaced, which typically means restarting the service or redeploying containers.

For environments where a patched build isn’t yet available through the normal update pipeline, temporarily disabling client-facing HTTP/2 on exposed listeners reduces exposure to both flaws, though it does not fix the underlying bugs. Anyone taking that step should verify from the client side that HTTP/2 is actually disabled, confirm sites and APIs still function correctly, and re-enable the protocol only after the real patch has been installed.

Share: Twitter  |  Facebook  |  LinkedIn
Join the discussion

This is a blog in the Fediverse: you can find this article everywhere with @blog@securebulletin.com and every comment/answer will appear here.

If you want to comment on Two Critical NGINX Flaws Put More Than 225 Million Sites at Risk of Pre-Auth Code Execution, use the discussion on Forum.

>> forum community

Comments

Leave a Reply