Skip to content
LIVE
The Executives BriefThe Executives Brief

AI found a 10-year-old NGINX flaw. 225M websites need the patch.

F5 shipped emergency fixes for 10 NGINX bugs, including two 9.2-rated remote code execution flaws, and researchers say AI is finding these holes faster than teams can patch them.

ByOmar Al-BalawiTechnology Correspondent, The Executives Brief
·4 min read
AI found a 10-year-old NGINX flaw. 225M websites need the patch.
Executive summary

F5 released emergency patches for 10 vulnerabilities in open source NGINX, including two critical 9.2-rated remote code execution bugs found with AI assistance by researchers at Wiz, Rapid7, Xint, and others. For executives, the disclosure signals that AI-driven bug hunting is surfacing decade-old flaws in foundational software faster than security teams can inventory, test, and deploy fixes.

F5 pushed emergency patches on Wednesday for 10 vulnerabilities in open source NGINX, the web server that routes traffic for roughly a third of the internet, and two of them are as bad as it gets. Tracked as CVE-2026-123633 and CVE-2026-123007, both carry a critical 9.2 CVSS rating and let an unauthenticated remote attacker execute code on a vulnerable server. The catch: HTTP/2 has to be enabled, which it is on a huge share of modern deployments. NGINX powers about 225 million websites, and F5 maintains the open source version. The company published an out-of-band security notification addressing the 10 CVEs. If you run NGINX, the instruction is blunt: update to 1.31.9 as soon as possible. One of the two critical bugs, CVE-2026-123007, affects versions dating back to NGINX 1.9.5 in 2015, meaning it sat in the code for at least a decade.

What makes this wave different is who found it. As with recently disclosed flaws in PHP, Apache, and Java, researchers leaned on AI to surface the NGINX bugs. Tomer Schwartz, Wiz's chief technology officer of exposure management and AppSec, helped find one of them alongside researchers from Ikotas Labs, Metnir, Rapid7, and Xint, who worked independently. All of those shops found at least one of the two critical CVEs, most if not all with an AI assist, and reported them to F5. Schwartz told The Register the wave of AI-discovered flaws "point to a new reality." His framing for defenders: "We need to treat every internet-connected system as potentially vulnerable." And: "AI has changed not only which vulnerabilities can be found, but how quickly longstanding risks can become real threats."

Both critical bugs are memory-safety failures in how NGINX handles HTTP/2 traffic. "In CVE-2026-123633, a connection's memory is freed while a request still needs it, creating a use-after-free," Schwartz and Wiz colleague Nir Ohfeld wrote in a Wednesday blog. "In CVE-2026-123007, incorrect flow-control accounting leads to an invalid memory copy." Rapid7 went further and published a proof-of-concept showing CVE-2026-123007 leveraged for unauthenticated remote code execution against a vulnerable linux/amd64 Docker image running nginx version 1.31.8. In plain English: a container that was current one patch cycle ago can be taken over from the network.

Xint found three more NGINX flaws plus a separate Linux kernel vulnerability that can be chained with any of the five NGINX bugs to gain root privileges on the host. The team named the six-bug set "nginx2root" and published videos showing CVE-2026-123007 used to execute commands as the NGINX worker inside a container, then the kernel bug used to escalate to root on the host. The attacks work on AWS, Microsoft Azure, Google Cloud Platform, and bare-metal servers. Xint said a researcher found CVE-2026-123007 while preparing for ZeroDay Cloud 2026, and "a few days later, a fully automated Xint scan found the other four without human guidance."

F5 says there is no sign yet that attackers are using any of this. "At this time, we have found no evidence of active exploitation of any of these vulnerabilities in the wild," an F5 spokesperson told The Register. "We continue to monitor closely and will share updates via myF5 as we learn more." That is the good news, and it is fragile. The same week brought a critical Apache flaw that is under active exploitation, which is the difference between a patch deadline and a fire drill.

The NGINX disclosure lands in the middle of a broader run of foundational-software scares: Tuesday's Java4Shell, found by AI bug-hunting outfit Calif, PHP2Shell, where Calif also used AI to find and weaponize critical PHP vulns, and that exploited Apache bug. Schwartz's read is that discovery is outrunning the machinery built to handle it. "AI is also accelerating vulnerability discovery faster than existing infrastructure was designed to absorb," he told The Register. "Even after a vendor ships a fix, security teams still face heavy manual work in terms of locating vulnerable versions and configurations, testing patches, deploying interim mitigations (such as disabling HTTP/2), and coordinating all of these rollouts." Translation: the patch is the easy part. Finding every instance is the hard part.

Schwartz does see a payoff down the road. "In the long run, AI-powered vulnerability research will harden foundational software," he said. In the meantime, he describes a nightmare for security teams, vulnerability managers, and system administrators drowning in near-daily CVEs they must patch before attackers get there. "Continued collaboration between security researchers, maintainers, and defenders is crucial to absorb the speed and scale of these disclosures," he said.

For anyone running infrastructure, the practical checklist writes itself from the source: confirm your NGINX version, get to 1.31.9, and know whether HTTP/2 is exposed on internet-facing servers. But the strategic lesson is bigger. If AI can find a 10-year-old memory bug in the world's most widely used web server, then every assumption that "we would have caught that" deserves a re-test. Asset inventories, container image freshness, and the speed at which a patch becomes a deployed fix are now board-level risk questions, not just tickets in a queue.

Executive ActionsLocked

This story's Key Insights and Take-aways are locked.

Create a free account to unlock Executive Actions for one credit.

Register to Unlock

Always free for Executives Club members. Join the Club

More in Technology