AI Swarm Attack Hits 395 Organizations via PaperCut Vulnerabilities

A Russian-speaking threat actor has deployed hundreds of autonomous AI agents to exploit vulnerabilities in PaperCut NG/MF print management software. The campaign targeted 440 instances across 395 organizations in 48 countries, demonstrating a dangerous evolution in the speed and scale of automated cyberattacks.

The attack leveraged two specific flaws: CVE-2026-81578, which allows for authentication bypass, and CVE-2026-82078, an unsafe reflection remote code execution (RCE) vulnerability. To power the swarm, the attacker integrated OpenAI's Codex harness and a DeepSeek model with a suite of penetration testing tools, including Mimikatz, Certipy, Rubeus, and Impacket.

According to confirmed data, the speed of the operation was unprecedented. Once the full campaign launched, 11 organizations were compromised within just 26 seconds. In one instance, a U.S. high school was breached from initial access to full domain administrator privileges in only seven minutes. The U.S. bore the brunt of the attack with 98 victims, and educational institutions were heavily targeted, accounting for 204 of the 440 compromised systems. Evidence shows the attacker first refined these exploits in a private lab environment that mirrored a vulnerable PaperCut deployment and an Active Directory server.

The Evolution of the Kill Chain

This incident represents a fundamental shift in the "cyber kill chain," the traditional model describing the stages of an attack from reconnaissance to payload delivery. Historically, these stages were executed sequentially, often over days or weeks. This campaign demonstrates the rise of "agentic attacks," where autonomous AI swarms execute these stages in parallel at machine speed.

GreyNoise researchers noted that large language models (LLMs) are enabling adversaries to move at a scale and velocity previously unseen. Cliff Steinhauer, director of information security and engagement at the National Cybersecurity Alliance (NCA), warned that hackers can now execute more robust attacks at scale, leaving defenders with significantly less time to react.

Industry Implications

This event signals a democratization of sophisticated attack capabilities. While high-speed, multi-step automation was once the exclusive domain of nation-states, the availability of open-weight models now allows a broader range of actors to automate complex workflows.

However, the breach also highlights that AI does not render traditional security obsolete. Experts emphasize that basic security hygiene—such as implementing multi-factor authentication (MFA), limiting administrative permissions, and utilizing Web Application Firewalls (WAFs)—remains an effective defense even against AI-driven swarms.

What to Watch

Security teams should monitor for similar agentic patterns where reconnaissance and exploitation happen almost simultaneously. As autonomous agents become more integrated with standard hacking toolsets, the window for detection and response will continue to shrink. Organizations using PaperCut software are urged to patch the identified CVEs immediately to prevent further autonomous exploitation.