Years-old Roundcube webmail flaw now under active exploitation, Canadian cyber agency warns

· CyberSecurity

A high-severity Roundcube webmail flaw patched back in May is now being actively exploited, Canada's Centre for Cyber Security warned this week, updating its original advisory after open-source reporting showed real-world attacks1. CVE-2026-48842 is a pre-authenticated SQL injection in Roundcube's virtuser_query plugin, letting an attacker bypass authentication and run malicious database commands to steal data without needing to trick a user into anything, though the attack itself is technically complex1. The flaw carries a CVSS score of 8.1 and stems from the plugin's preg_replace() filter, meant to neutralise injection via backslash escaping; attackers bypass it with crafted queries containing backslash sequences that defeat the regex escaping, letting malicious input concatenate quote characters into the SQL string sent to the database2.

Roundcube fixed the flaw in versions 1.6.16 and 1.7.1 back in May, but adoption has clearly lagged: Shadowserver currently tracks more than 523,000 internet-exposed Roundcube instances, though it is unclear how many remain unpatched versus already updated or run as honeypots1.

Why it matters: a four-month gap between patch and confirmed exploitation is a familiar pattern for widely deployed, self-hosted webmail software, and with over half a million instances exposed, unpatched Roundcube servers make an easy target for anyone scanning for the flaw.

Sources

  1. Hackers now exploit critical Roundcube flaw in code injection attacks (BleepingComputer)
  2. Roundcube Webmail Vulnerability in Attackers' Crosshairs (SecurityWeek)