Ad Code

Ticker

6/recent/ticker-posts

Sponsored by.

Chatbot AI, Voice AI and Employee AI. IndustryStandard.com - Become your own Boss!

Yehey.com - Java4Shell and Nginx2Root Flaws Put Internet Infrastructure at Risk

Image courtesy by QUE.com

In the space of about 24 hours, security researchers and vendors disclosed two critical vulnerabilities in the software that quietly runs much of the internet — and a third flaw, in Puppet Enterprise, was confirmed by the U.S. government to be under active attack. None of them requires a password. All of them have fixes available right now. The gap between "patched" and "exploited" is now measured in days, not months.

Java4Shell: A Bug in the Code That Decodes Your Login

The first bombshell landed on November 17, when Oracle released patches for CVE-2026-114938, a flaw researchers have dubbed "Java4Shell." It is a heap-based buffer overflow in the hardware-accelerated Base64 decoder used by OpenJDK and its downstream distributions, including Oracle Java SE and Oracle GraalVM. Oracle rates it High, with a CVSS v3.1 score of 8.1 and high attack complexity; the finders, security firm Calif, score it 9.2 under CVSS 4.0, according to Rapid7 and Tenable.

What makes Java4Shell dangerous is where the flaw sits. Java services decode Base64 data before they decide who you are — HTTP Basic credentials, JWT segments, and authentication tokens all pass through java.util.Base64 while the caller is still anonymous, as Harness explained in its technical write-up. A specially crafted Base64 string can make the fast decoder write past the end of a heap array, and Calif says it demonstrated unauthenticated remote code execution against several popular Java applications, including Atlassian Jira, Jenkins, and Keycloak.

The flaw only bites on a specific combination: OpenJDK 17.0.6 and later (with release lines 17, 21, 25, and 27 affected, per the OpenJDK Vulnerability Group) running on x86-64 processors that support AVX-512 VBMI, the CPU feature the fast decoder uses. Systems on ARM64, or x86-64 chips without AVX-512 VBMI, are not affected, according to Red Hat and Datadog Security Labs. Oracle credits Quang Luong of Calif.io and Albert Yang for reporting the vulnerability.

There is good news and bad news on exploitation. As of November 18, Tenable reported that GreyNoise had observed exploitation attempts against the flaw but had not confirmed any successful exploitation against its sensors, and researchers at Harness and Datadog found that reliable remote code execution requires additional work tailored to each application — denial of service is the more readily achievable impact. The bad news: Calif has published a local proof of concept, VulnCheck says a working exploit is available to its customers, and Kevin Breen of Immersive reports building one. Oracle's November 2026 Critical Security Patch Update covers 11 Java SE issues that may be exploitable over the network without credentials.

The fix is to move to Oracle's November replacement builds — 8u513, 11.0.33.1, 17.0.21.1, 21.0.13.1, 25.0.5.1, or 27.0.1.1, depending on the line. Until a fixed build is installed, starting the JVM with -XX:-UseBASE64Intrinsics disables the vulnerable decoder, and Datadog found that -XX:UseAVX=2 also blocks the vulnerable path on x86 deployments.

nginx2root: A Ten-Year-Old Flaw in a Third of the Web

If Java4Shell was the first shock, the second arrived on November 18, when F5 released security updates for five vulnerabilities in NGINX — including a critical unauthenticated remote code execution flaw that has existed in the code for more than a decade.

The lead flaw, CVE-2026-123007, is an integer underflow in NGINX's HTTP/2 flow-control accounting that leads to memory corruption and code execution with the privileges of the NGINX worker process. It was introduced when NGINX added HTTP/2 support in version 1.9.5, released in September 2015, and it remained in every official release through 1.31.8. F5 rates it 9.2 under CVSS 4.0 and 8.1 under CVSS 3.1. A second HTTP/2 flaw, CVE-2026-123633 — a use-after-free affecting releases 1.19.7 through 1.31.8 — carries the same 9.2 score.

The scale is staggering. NGINX is estimated to power about a third of the internet, according to CSO Online. Xint, one of the research teams that found the flaws, reported that as of November 10, Shodan returned 56.7 million hosts running NGINX, and of the 11.94 million with an identifiable version, 98.9% fell within the affected range. Those numbers are based on reported version strings and do not confirm a server is vulnerable — exploitation also requires HTTP/2 to be enabled, and a reported version may already include a backported fix — but the exposure surface is enormous.

The flaws were found independently by several research teams — Rapid7, Wiz, Xint, Ikotas Labs, and Metnir — and several of them used AI agents to find the bugs, adding to a string of recent AI-discovered vulnerabilities in widely used technologies such as PHP, Node, OpenSSH, and Java. Xint said its autonomous AI harness scanned 50,000 lines of code in 90 minutes to surface the critical flaw, and its CEO called it "essentially an internet breaker." Xint also found a separate, still-unpatched Linux kernel vulnerability in the SCTP subsystem, present since Linux 4.15 in 2018, that can be chained with any of the NGINX flaws to gain root privileges on the host. Together, Xint calls the collection nginx2root.

No in-the-wild exploitation has been publicly reported yet, and neither flaw is in CISA's Known Exploited Vulnerabilities catalog as of November 19, according to ThreatAft. But every research team built working exploits, and Rapid7 has published a public proof of concept — although it requires roughly 128 GB of HTTP/2 traffic and a specific glibc memory-copy implementation. Xint validated its attack over public IPs on AWS, GCP, and Azure under concurrent traffic.

The fix is NGINX 1.30.7 (stable) or 1.31.9 (mainline), and NGINX Plus 37.0.8.1 or 37.3.0.1. Where patching cannot happen immediately, disabling HTTP/2 blocks the two most severe flaws — but not the other three in the cluster, which sit in the resolver, HTTP/3, and mail proxy components.

Puppet Enterprise: The One Already Being Exploited

While Java4Shell and nginx2root are still in the "patch before someone weaponizes this" window, a third flaw has already crossed it. On November 17, CISA added CVE-2026-115731, a deserialization-of-untrusted-data vulnerability in Perforce Puppet Enterprise, to its Known Exploited Vulnerabilities catalog, citing evidence of active exploitation.

The flaw lets a client holding any CA-signed agent certificate submit a crafted report to a Puppet Server and execute arbitrary code as the service user. It scores 9.9 under CVSS v3.1 and 9.4 under CVSS v4.0. The catch, flagged by both CISA and Red Hat, is that where the certificate authority autosigns requests without strong verification, an unauthenticated network attacker can obtain a valid certificate and exploit the bug remotely. Perforce fixed it on November 3 in Puppet Enterprise 2023.8.13, 2025.12.1, and 2026.0.1 — but a working proof of concept has been public in the OpenVox advisory since that same day, so the two weeks between the fix and the KEV listing were two weeks with a copy-paste exploit available.

Federal civilian agencies have until December 1, 2026 to remediate under Binding Operational Directive 26-04. CISA has not said who is exploiting the flaw or whether ransomware crews are involved; its catalog lists ransomware campaign use as "Unknown."

AI Is Now Finding the Bugs — Faster Than Humans Can Patch Them

Step back from the individual CVEs and a pattern emerges. Java4Shell was found with help from AI coding tools, according to the CVETodo daily threat report. Four of the five NGINX flaws were found by Xint's automated harness. Security teams are now racing AI systems that can scan tens of thousands of lines of code in an hour and a half — and the defenders' patching window is shrinking accordingly.

CISA's acting director, Nick Andersen, told CyberScoop this week that the agency has identified roughly 100 critical-infrastructure companies for an initial round of security and resilience discussions, and CISA also published updated cybersecurity maturity guidance for dam operators on November 16. The message from Washington is consistent: the foundational software is under active scrutiny, by both researchers and attackers, and the basics — patching, inventory, and configuration — matter more than ever.

What to Do Today

First, find your Java. Inventory every service running OpenJDK 17.0.6 or later on x86-64 hardware with AVX-512 VBMI, especially internet-facing Jira, Jenkins, and Keycloak instances, and upgrade to Oracle's November builds. If you cannot patch immediately, start the JVM with -XX:-UseBASE64Intrinsics.

Second, find your NGINX. Check the version on every server and load balancer, upgrade to 1.30.7 or 1.31.9, and fully restart the service — a reload is not enough. If patching has to wait, disable HTTP/2 on exposed listeners, and remember that only blocks two of the five flaws.

Third, find your Puppet. If you run Puppet Enterprise, upgrade to 2023.8.13, 2025.12.1, or 2026.0.1 now — this one is confirmed exploited — and review your autosigning configuration before anything else, since open autosigning turns an authenticated flaw into an unauthenticated one.

Fourth, watch for the follow-on. GreyNoise is already logging Java4Shell probing, and the NGINX proof of concept is public. Expect scanning to spike within days, and treat unexpected crashes in Java services or NGINX workers as possible exploitation attempts, not random instability.

The Bottom Line

Two of the internet's most foundational technologies — Java and NGINX — each turned up a critical, unauthenticated remote code execution flaw within 24 hours, and a third flaw in Puppet Enterprise is already being exploited in the wild. Java4Shell hides in the code that decodes logins before authentication happens; nginx2root hid in plain sight for more than a decade, in software running on tens of millions of servers. The fixes for all three are already out. The only question left is whether your team patches before the scanning turns into exploitation — because this week, that window is measured in days.

Sources: Rapid7, "CVE-2026-114938: Reported Critical Pre-authentication Remote Code Execution in Java 17," November 2026; Tenable, "Java4Shell (CVE-2026-114938): Frequently asked questions about OpenJDK remote code execution vulnerability," updated November 18, 2026; Datadog Security Labs, "Java4Shell: Remote Code Execution in OpenJDK (CVE-2026-114938)," November 2026; Harness, "Past the End of the Buffer (CVE-2026-114938)," November 2026; Xint, "NGINX2Root: New Remote Code Execution Vulnerabilities in nginx Require Urgent Patching," November 2026; CSO Online, "AI unearths 'internet breaker' flaws in Nginx web server," November 18, 2026; SAFE Security, "NGINX HTTP/2 RCE: CVE-2026-123007 & CVE-2026-123633," November 18, 2026; ThreatAft, "nginx2root: 10-Year HTTP/2 Flaw Reaches Root," November 2026; Cyber Security News, "CISA Warns of Puppet Enterprise Vulnerability Actively Exploited in Attacks," November 18, 2026; CVETodo, "Daily Cyber Threat Report — 18 Nov 2026"; Security Boulevard, "Daily OT Security News: November 18, 2026."

Edited by Muse @QUE.COM

Website: https://QUE.COM Intelligence

Sponsored by: https://MAJ.COM AI Autonomous

Never miss a customer call again — MAJ.COM's AI receptionist answers every call, texts customers back instantly, and books the appointment for you. See how it works at https://MAJ.COM

Articles published by QUE.COM Intelligence via Yehey.com website.

Post a Comment

0 Comments

Comments

Ad Code