The Pulse
DIVD Says Its Hack May Have Used Agentic AI
The Dutch Institute for Vulnerability Disclosure says it was hacked and suspects the attack involved agentic AI. DIVD isolated its infrastructure and began a forensic investigation, but has not published technical evidence or confirmed the
AI.info Team · Written with AI ·
The Dutch Institute for Vulnerability Disclosure says it has been hacked and believes the attackers’ methods may point to agentic AI. DIVD has cut access to its infrastructure and started a forensic investigation with an outside incident-response team, but it has not published evidence establishing that an AI system carried out the attack.
“The modus operandi indicates that this is an agentic AI powered attack,” the nonprofit wrote in a Sept. 24 incident notice. DIVD said its investigation was ongoing and that it was treating the incident as a worst-case scenario, assuming a breach until proven otherwise.
DIVD isolates infrastructure as it investigates
The organisation detected suspicious activity, investigated it and concluded that its systems had been compromised. Its response includes blocking access to its infrastructure and bringing in a third-party team to help with forensics, according to DIVD’s statement.
DIVD said it had informed directly involved parties, reported the incident to the Dutch data-protection authority and the National Cyber Security Centre, and discussed its options with police. It listed infrastructure isolation and security, forensic work, and support for its volunteers among its immediate priorities.
The nonprofit did not identify an entry point, affected systems, an attacker, a specific AI model or tool, or any confirmed data theft in that notice. Its public statement describes an assessment of the attackers’ methods, not a technical report demonstrating how the intrusion happened.
AI attribution remains an early assessment
A Sept. 28 analysis by Breachroad draws a line between DIVD’s confirmed disclosure of a compromise and its interpretation of the attack. The article says the initial notice does not establish that a particular autonomous system was used, and notes that the post does not identify an actor or a specific AI tool.
That distinction matters: a claim about an attack’s likely method is not the same as proof that an AI agent directed it. DIVD says its investigation is still under way, and the evidence needed to assess the AI claim has not been made public.
Breachroad also notes that the initial notice does not state the incident’s full scope, including the path into DIVD’s systems, what data may have been available or whether information was exfiltrated. The nonprofit’s “assume breach” language describes a cautious incident-response stance; it does not establish that every dataset was taken.
A sensitive target for a vulnerability-disclosure group
DIVD’s volunteers find exposed systems and notify their owners, according to the nonprofit’s description cited by Breachroad. A compromise at an organisation doing that work raises questions about whether attackers could have accessed information connected to vulnerability research or notifications, but the published notices do not say that such information was reached or stolen.
The incident also puts pressure on the distinction between identifying automated activity and establishing its cause. DIVD has said what it suspects and what it is doing to contain and investigate the breach; it has not yet published the technical details that would show how the attackers gained access or substantiate the agentic-AI assessment.
DIVD’s Sept. 24 notice said it expected to provide another public update on Monday, Sept. 28, or sooner if it had more to share. The key unresolved facts are the intrusion path, the systems and data affected, and what evidence supports the suspected use of agentic AI.