The Dutch Institute for Vulnerability Disclosure (DIVD), a nonprofit dedicated to identifying security flaws, disclosed that its own systems were compromised on September 21, 2026, in what it described as the first breach in its seven-year history.
The attack leveraged chained zero-day vulnerabilities in DIVD's Zammad ticketing system to achieve session hijacking, code execution, and root-level privilege escalation within seconds. The attacker stole researcher emails during the intrusion.
DIVD disclosed the incident on LinkedIn, characterizing the attack's operational pattern as agentic AI—a method the organization had not previously encountered. The exploitation of two previously unknown flaws in Zammad exposes a critical supply chain risk for organizations relying on third-party ticketing platforms for operational security.
Following discovery, DIVD immediately isolated its compromised infrastructure and initiated a forensic investigation with a third-party incident response firm. The organization notified the Autoriteit Persoonsgegevens (Dutch data protection authority), the National Cyber Security Centre, and local police.
DIVD is operating under a worst-case scenario, assuming total system compromise pending forensic results. Current priorities include maintaining infrastructure isolation, completing forensic analysis, and supporting affected volunteers.
The organization described the event as "loud and very, very messy." Named entities Storm-3168 and JADEPUFFER have been linked to the autonomous AI agent, though attribution remains under investigation. DIVD committed to a public update by September 28.
