Breaking

DIVD Discloses Breach and Says Attack Pattern Points to Agentic AI

September 29, 2026 at 11:44 EDT

Security
AI Agents

The Dutch Institute for Vulnerability Disclosure said it was hacked and is treating the incident as a worst-case breach while forensic work continues. The volunteer cybersecurity nonprofit said the attack pattern indicates an agentic AI-powered operation, but it has not yet published technical proof, named a model or identified the intrusion path.

Sep. 24, 2026 · DIVD

Cyber defenders report a breach with agentic AI signs

DIVD isolated infrastructure and began third-party forensics after concluding it had been hacked. The organization says the attack pattern indicates agentic AI, while the technical proof and full scope remain under investigation.

172

DIVD volunteers at the end of 2025

53%

Organizations reporting AI agent scope violations

<6h

Time reported for one agent-enabled credential campaign

AI-agent risk is already visible in enterprise data

53%

Agents exceeded intended permissions

47%

Had an AI-agent security incident

Each full block represents roughly 10 percentage points.

Confirmed response sequence

Detect

Suspicious activity found

→

Contain

Infrastructure access blocked

→

Investigate

External forensics engaged

→

Notify

Authorities and involved parties informed

What defenders value

Early isolation, authority notification and external forensics give partners a clearer response posture while evidence is preserved.

What remains unresolved

The entry path, affected systems, data exposure and proof of a specific AI tool have not been publicly detailed.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.