Threat level Guarded
Submit a tip

Major threats, straight to your inbox

Only the threats that matter: actively exploited flaws, big breaches and attacks that need action. No spam, unsubscribe any time.

UnrealIRCd CVE-2026-114713 Exploited: Label Flaw Grants Operator Power

UnrealIRCd says attackers are exploiting CVE-2026-114713, an unescaped IRCv3 label in the labeled-response module that lets ordinary users inject server-to-server commands such as KILL and GLINE on linked networks. Version 6.2.7.1 and hot-patch bundle-20261107 fix it.

UnrealIRCd CVE-2026-114713 Exploited: Label Flaw Grants Operator Power

Threat card

Status
Actively exploited
Threat type
Patch Watch
Disclosed
Nov 7, 2026
Affected
UnrealIRCd 6.0.0 to 6.2.7 on networks with two or more linked servers; end-of-life UnrealIRCd 5 is similarly affected

The short version

  • CVE-2026-114713 lets ordinary IRC users inject commands that linked servers run with server authority.
  • The flaw sits in the labeled-response module, which echoes the IRCv3 label tag into replies without escaping it.
  • The UnrealIRCd team calls it the most critical issue in 15 years and confirms live exploitation; CISA lists it in KEV.
  • Upgrade to 6.2.7.1 or load hot-patch bundle-20261107 on every server, with a spamfilter rule as a stopgap.

The UnrealIRCd project says attackers are actively exploiting CVE-2026-114713, a flaw in the IRC server’s labeled-response module that lets any ordinary user run commands with server authority across a linked network. The UnrealIRCd CVE-2026-114713 bug carries a CVSS score of 8.6, affects versions 6.0.0 through 6.2.7, and CISA listed it as a known exploited vulnerability on 13 November 2026. Network operators should load hot-patch bundle-20261107 or upgrade to 6.2.7.1 on every server today, because one unpatched hub is enough to expose the whole network.

What happened

CVE-2026-114713 is a command injection flaw in UnrealIRCd’s labeled-response module. According to SecurityOnline’s summary of the project advisory, the module does not escape the IRCv3 label message tag when it echoes that tag into replies, so an attacker can smuggle extra lines into a reply that linked servers treat as server-to-server commands.

The UnrealIRCd team calls it “the most critical issue found in UnrealIRCd in the past 15 years.” It says the bug was found during an internal audit “with the help of AI” and, in an updated forum warning, that “bad guys” are exploiting it on the internet right now.

Timeline

Date Event
1 July 2023 UnrealIRCd 5 reaches end of life; it is similarly affected
7 November 2026 Date of hot-patch bundle-20261107 for 6.0.0 to 6.2.7; UnrealIRCd 6.2.7.1 also carries the fix
9 November 2026 Public reporting on the flaw; project later confirms live attacks
13 November 2026 CVE-2026-114713 listed in the CISA KEV catalog

How the attack works

Root cause

IRCv3 message tags let a client attach a label to a command so it can match the server’s reply. UnrealIRCd’s labeled-response module copies that label into its replies without escaping control characters. The weakness maps to CWE-140, improper neutralization of delimiters, which is how an attacker breaks one reply into several protocol lines.

The injected lines become dangerous once they reach another server. In a network with two or more linked servers, the receiving server reads them as server-to-server traffic and runs them with server authority. The project says “any ordinary user can KILL, GLINE, and do many more bad things,” which means disconnecting users and banning them network-wide without operator rights.

Preconditions

The attack needs only a normal client connection and a network of at least two linked servers. Standalone servers are not exposed to the cross-server path, but most public and community IRC networks are linked by design.

MITRE ATT&CK mapping

Tactic Technique (ID) How it was used here
Initial Access Exploit Public-Facing Application (T1190) Crafted label tags sent by an ordinary IRC client
Privilege Escalation Exploitation for Privilege Escalation (T1068) Injected lines run with server authority on linked servers
Impact Account Access Removal (T1531) KILL and GLINE used to disconnect and ban users

Who is behind it

Not yet attributed. The UnrealIRCd team has confirmed exploitation but has not described the attackers, their infrastructure or the networks hit. No public reporting ties the activity to a named group.

Affected products and versions

Product Affected Fixed CVE CVSS
UnrealIRCd 6 6.0.0 to 6.2.7 6.2.7.1 or hot-patch bundle-20261107 (no restart) CVE-2026-114713 8.6 (v3)
UnrealIRCd 5 All, end of life No fix; blacklist module and migrate CVE-2026-114713 8.6 (v3)
UnrealIRCd 3.2.x and 4 Not checked by the project Not applicable CVE-2026-114713 Not disclosed

Version 6.2.7.1 also fixes two IP address leaks, one of which allows decloaking users, plus crashes in CHGIDENT and whowasdb, a blacklist recheck hang, a connection counting bug and a TLS renegotiation flood.

Indicators of compromise

No IOCs have been published by the project or CISA. Behavioural signs are the useful signal: KILL, GLINE or other operator-level actions that no IRC operator issued, client commands carrying unusually long or malformed label tags, and server notices showing commands attributed to a linked server at the same moment a regular user sent a labeled command.

Detection and hunting

Review UnrealIRCd logs on every server since early November for KILL and GLINE events and match each one against the list of operators who were online. Unexplained network bans or mass disconnects are the most visible symptom. Enable logging of spamfilter hits once the project’s invalid-label rule is in place, since each hit is a likely exploitation attempt.

Then confirm patch status on every node: the labeled-response module should report bundle-20261107 or version 6.2.7.1. Our N-central hotfix coverage shows why partial rollouts across linked systems leave the weakest node as the entry point.

Mitigation and response

  1. Run hot-patch bundle-20261107 on every server from 6.0.0 to 6.2.7, or upgrade to 6.2.7.1 and restart.
  2. Verify the patch level of the labeled-response module on each server after the update.
  3. Until every node is patched, add the project’s spamfilter rule blocking invalid labels; it protects linked servers on 6.0.4 or newer.
  4. On UnrealIRCd 5, blacklist the labeled-response module, restart and schedule migration to version 6.
  5. Audit and lift bans and kills that no operator issued, and review operator and services accounts for unexpected changes.

Analyst take

IRC looks like legacy technology, but it still runs open-source project channels, gaming communities and some operational chat. A flaw that lets any user act like a server is a full loss of control for the network, and the project’s own “15 years” framing should be read as a priority signal.

Two trends stand out. The bug was found by an AI-assisted audit, and attackers moved on it within days of the fix. Maintainers of small protocol servers should expect the same compression of the patch window that enterprise software already faces, and should ship hot-patch mechanisms like UnrealIRCd’s no-restart bundle wherever they can.

What to do now

0 of 5 done

Questions readers ask

What is CVE-2026-114713?

It is a flaw in UnrealIRCd's labeled-response module that fails to escape the IRCv3 label tag, letting users inject commands that linked servers execute.

Which UnrealIRCd versions are affected?

Versions 6.0.0 through 6.2.7 are affected, and the end-of-life UnrealIRCd 5 branch is similarly affected.

How do I fix CVE-2026-114713?

Upgrade to UnrealIRCd 6.2.7.1 or run hot-patch bundle-20261107, which needs no restart, on every server in the network.

Sources

  1. UnrealIRCd forums: security warning for CVE-2026-114713forums.unrealircd.org
  2. SecurityOnline: UnrealIRCd vulnerability CVE-2026-114713 exploitedsecurityonline.info
  3. CISA: Known Exploited Vulnerabilities Catalogcisa.gov
Share this story
Get the next major threat first

One short email when something serious breaks.

Threatspedia Desk

Threatspedia covers major cyber threats only, verified against primary sources. Read our editorial policy, or report a correction.

Threat alerts

Be first to know when a major threat breaks.

One short email when something serious happens, plus a weekly roundup. Free, no noise.