Dutch Cybersecurity Non-Profit Compromised in Agentic AI Attack
The Dutch Institute for Vulnerability Disclosure (DIVD), a non-profit organization dedicated to the ethical disclosure of vulnerabilities in digital systems, has revealed a recent security breach that highlights the increasing threat posed by sophisticated cyber attacks involving artificial intelligence. Having established its reputation for making the digital world a safer place, DIVD found itself a target after noticing suspicious activities on its network on September 24.
In a LinkedIn post dated September 30, the organization disclosed that hackers exploited two zero-day vulnerabilities in its helpdesk platform, Zammad. These vulnerabilities allowed the assailants to hijack user sessions, execute code remotely, and escalate their privileges from a standard Zammad user to root access almost instantaneously. This rapid escalation enabled them to delve deeper into DIVD’s systems, access various other services, and exfiltrate sensitive data. The organization urgently advised all users of Zammad to upgrade to version 7 promptly or take their systems offline to mitigate the risk of similar attacks.
The flaws exploited during this incident have been identified as remote code execution vulnerability CVE-2026-102489 and an elevation of privileges flaw CVE-2026-102490. Both vulnerabilities were assigned a Common Vulnerability Scoring System (CVSS) score of 9.4 when exploited together, making them particularly dangerous.
Despite the severity of the breach, DIVD’s expertise in cybersecurity allowed it to contain the situation effectively. The organization reported that proper network segmentation and the proactive actions of its IT and incident response teams played crucial roles in halting the attackers’ advance into more sensitive areas of its network. However, as the organization admitted, some degree of damage had already occurred before containment measures were implemented.
A subsequent examination of the incident revealed that personal information belonging to volunteers, including their email addresses and possibly other contact details, had been compromised. This breach raises concerns about potential phishing attempts, as malicious actors might exploit the stolen information to impersonate DIVD staff.
In the face of these challenges, DIVD’s investigation unveiled the remarkable role AI played in the attack. Log analysis indicated that the attackers used AI scripts that contained notes justifying their actions. These notes elaborated on why their activities were legitimate, a tactic that a typical human attacker would not typically employ. The insights shared by DIVD strongly suggest that the breach was a product of an agentic AI-powered attack, raising alarms about the future implications of such technology in malicious activities.
Tim Burke, CEO of Quest Technology Management, emphasized the urgency of addressing these emerging threats. He noted that AI-driven attacks tend to compress the timelines for detection and response, which can leave organizations vulnerable, especially those without dedicated Security Operations Centers (SOCs). Burke remarked that organizations need continuous monitoring and visibility to swiftly address any anomalies in their environments.
"The fundamental principles of cybersecurity are more crucial than ever," he explained. "While AI can assist in various areas, it does not replace the foundational practices of patching, monitoring, access control, immutable data storage, segmentation, and incident response. Each of these elements must remain a priority."
Burke also highlighted the importance of network segmentation in containing damage during cyber incidents. He stressed that immediate containment efforts are vital in the first hour following detection, recommending actions such as isolating affected systems, restricting access from compromised accounts, blocking suspicious connections, and preventing further movements within the network.
He further emphasized the significance of pre-established authority regarding containment measures within organizations. "In an age where attack activity can transpire at machine speed, delays become particularly consequential. It’s critical for organizations to know who is responsible for taking immediate action."
This incident underscores the pressing need for organizations to enhance their cybersecurity infrastructures and preparedness to counteract the ever-evolving tactics employed by cybercriminals, especially those leveraging advanced technologies like AI. As cyber threats grow more sophisticated and challenging to detect, the discourse surrounding responsible AI and its implications in both offensive and defensive strategies becomes increasingly vital. Keeping digital environments secure will require persistent vigilance, innovation, and a commitment to effective incident response strategies.