Attackers Are Exploiting a Critical Flaw in SiteGuard's Web Firewall. Here's What to Know.

Key points

• Attackers are actively exploiting CVE-2026-106931, a command injection flaw in SiteGuard Server Edition that lets them run commands as root or SYSTEM, the highest level of access on a server.
• The bug sits in the product's statistics feature, which is turned on by default.
• Fixed versions are SiteGuard Server Edition 8.00 Update 5 for Apache and Nginx builds, and 7.00 Update 6 for FreeBSD.
• Paying support customers were told about the flaw on September 29, 2026, before the public advisory.
• SiteGuard warns that applying the patch may erase forensic traces, and tells administrators to download diagnostic information first.

Hackers are breaking into servers running SiteGuard Server Edition, a web application firewall sold by a major Japanese security vendor, by abusing a flaw that hands them full control of the host machine.

A web application firewall sits in front of a website and filters out malicious traffic before it reaches the site. SiteGuard says its products protect more than 1.5 million websites, mostly in Japan.

The flaw, tracked as CVE-2026-106931, is an OS command injection bug: an attacker can smuggle their own system commands into a feature that was only meant to process data. Those commands run with root privileges on Linux or SYSTEM privileges on Windows, which is the keys-to-the-building level of access. The weakness lives in the statistics function, enabled by default on new installs, so most customers are exposed unless they turned it off.

Which versions need patching?

Upgrade to SiteGuard Server Edition 8.00 Update 5 on Apache httpd or Nginx (including IIS, non-FreeBSD), or to 7.00 Update 6 on FreeBSD builds. IIS editions from 3.10 through 4.00 Update 4 are not affected, according to JPCERT/CC, Japan's national computer emergency response team.

For installs that can't be upgraded cleanly, SiteGuard is shipping a file-replacement patch covering the statistics code, including for end-of-life builds.

What should administrators do first?

Preserve the evidence before you patch. The vendor's own advisory warns that applying the fix can overwrite traces of an attacker's activity, which is unusual language in a security bulletin and worth taking at face value.

SiteGuard tells administrators to pull a diagnostic bundle from the web management console (Web management screen, then Diagnostic information, then Download diagnostic information) and keep it before touching anything else. That file is what an incident responder would need later to work out whether the server was already breached.

Paying support customers were notified on September 29, 2026. The public advisory came later, meaning sites without a direct support contract were in the dark while exploitation was already underway. We've tracked four command-injection cases in the last 90 days, and this one stands apart because the vulnerable component is the security layer itself, not the application it's meant to protect.

Who is behind the attacks?

Nobody's saying yet. Neither SiteGuard's bulletin nor the JPCERT/CC warning names an attacker, lists victims, or describes how exploits are being delivered. BleepingComputer, which flagged the active exploitation, noted the same gap. JPCERT/CC last appeared in our coverage on 8 October, when the agency was warning about attackers hitting known software flaws across Japanese infrastructure.

The honest read here: a flaw in a security product carries a different weight than a flaw in an ordinary web app, because the firewall is usually the thing standing between the internet and everything else. If the statistics endpoint on your SiteGuard box is reachable from the public internet and you haven't patched, treat it as already compromised until you can prove otherwise.

There's also a separate SiteGuard WP Plugin for WordPress sites from the same vendor. That plugin is a different product and isn't what this advisory covers.