Attackers are abusing a command injection bug in SiteGuard Server Edition's statistics feature. Here is who is exposed, what the patch does, and why you should save evidence first.
Attackers are exploiting CVE-2026-106931, a command injection bug in SiteGuard Server Edition, a Japanese web application firewall. The flaw sits in the statistics feature, which is on by default, and can let an attacker run commands as root or SYSTEM.
The vendor offers fixed releases and a file patch for older versions, and turning off statistics works as a stopgap. It warns that patching can erase attack traces, so admins should download the diagnostic information from the management screen first. The bug is rated as needing some prior access, so attackers likely chain it with stolen logins or another weakness. Who is behind the attacks is still unknown.
A tool that exists to block attacks on websites has become the way in. Attackers are exploiting a flaw in SiteGuard Server Edition, a Japanese web application firewall, that can let them run commands with the highest level of access on the server it is meant to protect.
Most people will never touch SiteGuard directly, but plenty of websites they use sit behind it. The vendor says its products protect more than 1.5 million websites. If a firewall like this is taken over, the attacker can change settings, plant files or read data on the web server itself. For site owners, admins and anyone who pays a hosting company to run their site, this is a patch-today problem, with one twist: you should save evidence before you fix it.
The SiteGuard WAF vulnerability, tracked as CVE-2026-106931, is an OS command injection bug. In plain terms, data that the software processes can be crafted so that it slips operating system commands through, and the server then runs them. A WAF, or web application firewall, is software that inspects traffic heading to a website and blocks attacks such as SQL injection or malicious scripts.
The weak spot is the statistics function, the part of SiteGuard that tallies detected attacks by type, source address and time. It is switched on by default. When the flaw is exploited, commands run as root on Linux and Unix systems or as SYSTEM on Windows, the most powerful accounts on each.
The developer, EG Secure Solutions, says abuse has already been confirmed. Japan’s CERT coordination center issued its own warning, and the bug carries a severity score of 8.5 under the current CVSS 4.0 scale (7.8 under the older version), which counts as high. Both the vendor and Japan’s vulnerability database labeled the alert urgent.
Only SiteGuard Server Edition installs that have the statistics function turned on are at risk. The affected releases are:
One common mix-up is worth clearing up. There is also a popular WordPress security plugin called SiteGuard, but it is a separate product and is not the one covered by this warning. The vulnerable software is the server-side firewall that plugs into the web server.
The vendor’s scoring describes the attack as local and requiring low privileges. That means the bug is not rated as something a stranger can trigger from anywhere on the internet with no foothold. The likely path is that an attacker first gets limited access to the server, through stolen credentials or another weakness, then uses the statistics flaw to jump to full control.
That is a description of what is possible, not a confirmed account of these attacks. Neither the vendor nor Japan’s CERT has said who is behind them, who was hit, how many systems were affected or what tools were used. The vendor told customers with direct support contracts about the problem in late September, before going public in mid-October.
The US government has taken notice too. The Cybersecurity and Infrastructure Security Agency added CVE-2026-106931 to its Known Exploited Vulnerabilities catalog, alongside the Plex2Shell bugs in Plex Media Server and a Siemens Software Center flaw, and told federal civilian agencies to fix it by October 23, 2026.
The order of steps matters here, because the vendor warns that applying the fix can wipe out traces of an attack. A sensible sequence looks like this:
If you suspect an attack is happening right now, don’t hold off on blocking it just to preserve evidence. Do both in parallel with whoever manages the server.
For most readers, the risk is indirect. You won’t see a pop-up or a warning; the danger is that a site you trust, an online shop or a company portal, could be quietly taken over and used to steal data or serve malware. That is why the patch window matters more than the headline.
For small businesses, the useful takeaway is to know which security products sit on your server and who is responsible for updating them. A firewall that nobody patches can turn from shield into weak point.
The vendor’s fixed builds and patches are available now, and federal agencies were already past their October 23 deadline, so any SiteGuard server still on an older release should be treated as overdue.
It is an OS command injection flaw in the statistics feature of SiteGuard Server Edition that can let an attacker run commands as root or SYSTEM. The vendor confirmed it is being exploited.
No. The WordPress plugin with the SiteGuard name is a different product. The warning covers SiteGuard Server Edition, the firewall that runs inside the web server.
Download the diagnostic information first, then update to 8.00 Update 5 (or 7.00 Update 6 on FreeBSD) or apply the vendor’s file patch. Turning off the statistics function is a short-term workaround.
You're reading the quick version.
Read next
WordPress vulnerability today: is your site on 7.1.4 yet?
A race condition in the WordPress update process is under attack, CISA confirms. Here is who is exposed, what the fix changes and what to check.
Attackers are locking files on outdated Synology NAS boxes with the .L0CK3D extension. Here are the DSM builds that fix it and the steps Synology recommends.
Russia-linked hackers paired an old Chrome bug with a Qualcomm graphics flaw to rob Android phones. Two quick checks show whether yours was ever exposed.
An unreleased OpenAI model found and exploited hundreds of Linux kernel bugs for about $1,279 each. Here's what's patched, who's exposed and what to do.
The Daily Brief
Every morning, the news that matters from AI, phones, apps and the people shaping tech. Explained in plain English. Free.
One email a day. No spam, unsubscribe anytime. Privacy policy
