MediumII Cyber Warfare & Critical Infrastructure5 October 2026, Monday
Zammad flaws exploited by an AI agent in the Netherlands join CISA's list
The US Cybersecurity and Infrastructure Security Agency (CISA) added two flaws in Zammad helpdesk software to its Known Exploited Vulnerabilities (KEV) catalogue on 5 October. They had been chained on 21 September in an attack, attributed to an AI agent, on the Dutch disclosure body DIVD.
According to Cyberpress, CISA added CVE-2026-102489 and CVE-2026-102490 to the KEV catalogue on 5 October. Under directive BOD 26-04, federal civilian agencies must prioritise patching vulnerabilities in the catalogue. The first flaw allows unauthenticated session hijacking and remote code execution in versions 6.3.0–6.5.4. The second lets a low-privileged user escalate to root. Help Net Security reports that the second flaw affects all versions, and DIVD advises users to upgrade to version 7 or take systems offline.
According to the timeline reported by SecurityOnline, the attack began on 21 September and was detected on 22 September. DIVD reported the flaws to Zammad on 24 September, began scanning internet-facing installations on 26 September and disclosed the entry point publicly on 30 September. The attackers stole email addresses and some contact details of DIVD volunteers. Network segmentation halted further penetration after 22 September. Sources diverge on severity: SecurityOnline scores the two flaws at 8.7 and 8.5 and the chain at 9.4, while Cyberpress rates the first flaw alone at 9.4.
DIVD's records describe the attacker as an autonomous AI agent that justified its own actions and chained the two flaws to reach root within seconds. That the target was itself a vulnerability disclosure organisation shows helpdesk software to be a second-tier attack surface, including at critical infrastructure operators. The number of affected installations had not been disclosed as of 7 October. The identity of the actor behind the attack could not be verified as of 7 October.
Talay assessment
Bottom line
The significance lies less in Zammad itself than in the shape of the attack. An autonomous AI agent chaining two zero-days to reach root within seconds shows the patching window narrowing further than it would against human attackers. CISA's decision to add the flaws to KEV on 5 October signals that the exploitation risk is being taken seriously on US government networks too.
Likely effects
- Corporate patching burdenNegativeWeeks
Because the root escalation flaw affects all versions, upgrading to version 7 may not suffice; organisations may have to pull helpdesk systems off the internet.
- Cyber insuranceNegative1–6 months
Faster attacks attributed to AI agents could lead insurers to tighten their requirements on patching times and network segmentation.
- TürkiyeUncertainWeeks
Turkish public and private bodies using open-source helpdesk software are exposed to the same two flaws; the number of affected local installations could not be verified.
Possibilities, ranked
- 1Contained spread55%
Organisations limit the risk through upgrades and taking systems offline; few new victims are disclosed.
Watch: Zammad releasing a full patch for the root escalation flaw
- 2Mass exploitation35%
The chain becomes public exploit code and several organisations in Europe and the US report data breaches.
Watch: Zammad-related victim listings on ransomware leak sites
- 3Actor identified10%
The attack is attributed to a specific state-linked or criminal group, giving the incident a political dimension.
Watch: An attribution statement from Dutch authorities or DIVD
Probabilities are calibrated judgement based on the sources, not measurement, and are revised as new information arrives. Not investment advice.
Market reaction
Indicators affected
- Zammad flaws added to KEV▼ 2
- Chain severity score (CVSS)▼ 9.4
- Attack to public disclosure▼ 9 days