A newly uncovered AI-powered attack on Papercut print management software has sent shockwaves through the cybersecurity community, highlighting the rapid evolution of cyberattack strategies. In a matter of hours, an AI swarm compromised over 440 instances of Papercut NG and MF hosted by 395 organizations in 48 countries, demonstrating the potential for unprecedented speed and scale in cyberattacks.
The attack, attributed to a likely Russian-speaking actor, leveraged hundreds of AI agents trained in a lab environment to seek out vulnerable instances of Papercut. The attackers exploited two previously unknown vulnerabilities in the software, using them to gain remote code execution (RCE) capabilities. What’s more, they managed to compromise Windows Active Directory environments and opportunistically attempt to spread their influence.
This incident is significant not only because of its scope but also because it showcases the increasing role of AI agents in every stage of the cyber kill chain. The cyber kill chain model outlines the steps attackers take to move from reconnaissance to compromise to delivering a payload, and AI agents are now being integrated into each of these stages. This development marks a major shift in the way attacks are carried out, enabling adversaries to operate with greater speed and efficiency.
The ease with which attackers can access large language models (LLMs) has democratized their use, allowing even less sophisticated threat actors to incorporate AI capabilities into their attacks. “While nation-state actors may have a head start in terms of access to commercially available AI tooling and training,” says Kelli Vanderlee, senior manager at the Google Threat Intelligence Group (GTIG), “the willingness to experiment with these capabilities is becoming increasingly widespread among all types of threat actors.”
Beyond the Papercut incident, there are signs that attackers are using AI agents not only for reconnaissance but also for lateral movement and exfiltration. The fact that they can now create lab environments for staging and testing agentic attacks raises concerns about the potential for fully autonomous cyberattacks.
As Cliff Steinhauer, director of information security and engagement at the National Cybersecurity Alliance (NCA), notes, “It’s impossible to responsibly put an exact timeline on fully autonomous attacks. However, we don’t need to wait for that to happen to know AI is already changing the game.” With hackers able to execute more robust attacks at scale and defenders facing reduced time to react, it’s essential for organizations to stay vigilant and adapt their security strategies to address these evolving threats.
To mitigate the risks associated with AI-augmented cyberattacks, it’s crucial for defenders to adopt a proactive approach. This includes staying informed about emerging attack techniques, continuously monitoring for signs of AI-powered activity, and implementing robust incident response plans. Moreover, organizations should prioritize training and upskilling their security teams to address the growing threat landscape. By doing so, they can better prepare themselves against the rapidly evolving threats posed by AI-powered attacks.
Source: Dark Reading — 2026-09-11