AI-Assisted PaperCut

A suspected Russian-speaking cyber actor has been using AI to develop exploits for a pair of PaperCut vulnerabilities. The result? Hundreds of compromised systems in a matter of weeks.

According to The Hacker News, independent reports from Blackpoint Cyber and GreyNoise traced the activity back to an IP address 45.142.193[.]132 that's been linked to unauthorized port scanning and brute-force attempts. Arctic Wolf also flagged the same address in connection with exploitation activity last week.

The campaign targets CVE-2026-81578 and CVE-2026-82078, a combination of authentication bypass and remote code execution. The primary victims are in the education sector across the U.S., U.K., France, Spain, Canada, Belgium, Portugal, Australia, Germany, and Switzerland.

Let me break down what's happening.

Quick Summary

What Is PaperCut?

PaperCut NG and MF are print management software used by organizations around the world. It is used by schools, universities, and businesses for tracking printing, quotas, and cost management purposes.

It's not the kind of software you'd expect to be a major target. But that's exactly what makes it attractive. It is widely used, facing the Internet, but often neglected by the security team.

The two vulnerabilities exploited in this campaign CVE-2026-81578 and CVE-2026-82078 are a dangerous combination. One bypasses authentication. The other allows remote code execution. Together, they give an attacker complete control.

The AI Connection

This is where things get interesting. The attacker didn't just use AI to write a few scripts. They built an entire exploitation pipeline around it.

According to Blackpoint, the earliest activity began on August 31. The attacker started by comparing patched and unpatched PaperCut builds to understand the vulnerabilities. Within hours, that research had been turned into a multi-threaded validation tool.

The Workflow:

• Vulnerability research using AI-assisted analysis
• Exploit development
• Target filtering and country exclusion
• Automated exploitation
• Failure analysis
• Code changes and retry waves

The attacker used OpenAI Codex and a DeepSeek model to power hundreds of AI agents. These agents ran offensive security tools like Mimikatz, SharpHound, Certipy, Rubeus, and Impacket.

Two Open-Source Tools Supported the System:

• Hindsight: Provides persistent memory for AI agents
• AionUi: A graphical workspace to run multiple AI agents concurrently

"The strongest AI impact in this campaign was not a novel exploit technique," Blackpoint said. "It was the reduction of human effort required to research, develop, debug, classify, track, retry, and continuously improve exploitation across hundreds of real systems."

That's the key insight. AI didn't make the exploit more powerful. It made the attack more efficient.

The Scale of the Attack

The numbers are staggering.

GreyNoise said the attacker compromised at least 11 organizations in 26 seconds once the campaign was running. In one attack targeting a U.S. high school, the time between initial access and full domain administrator access was just seven minutes.

The Timeline:

• Empty workspace to first RCE against a real victim: under 4 hours
• 11 organizations compromised: 26 seconds
• High school initial access to domain admin: 7 minutes

That's not a typo. Seven minutes to full domain admin.

How the Attack Worked

Step 1: Reconnaissance

The attacker used Netlas.io, an internet scanning service, to build target lists. They found internet-facing PaperCut systems and filtered them by country.

Step 2: Exclusion Policy

The attacker tried to avoid targeting entities in 28 countries, including Russia, China, Hong Kong, Thailand, Iran, Venezuela, Indonesia, Pakistan, and Bangladesh. But GreyNoise noted that the "attempted restraint failed in some instances."

Step 3: Exploitation

Exploitation took place using the authentication bypass vulnerability and the RCE vulnerability. AI agents were employed to run post-exploitation tools.

Step 4: Post-Exploitation

Arctic Wolf observed the delivery of Windows registry hive collection tools, Metasploit/Meterpreter-related Java payloads, and commands used to identify hosts, users, processes, and sensitive configuration data.

Step 5: Target Classification

The attacker classified targets based on operating systems, environments, and states:

• Active
• Unreachable
• Lacking certain phases
• Post-exploitation candidates
• Waiting for retry

It’s not a one-size-fits-all approach. It’s a highly intelligent pipeline capable of dealing with failures.

What We Don't Know

The End Goal:

"At this time, we cannot confirm the exact end goal of this campaign," said Nevan Beal, principal MDR analyst at Blackpoint. "The threat actor's methodology is consistent with initial-access activity, but we do not yet have sufficient evidence to confirm whether they are operating as an initial access broker."

The Attribution:

The actor is suspected to be Russian-speaking, but no specific group has been named.

The Full Scope:

"There are other real victims that could not be attributed to a named organization," GreyNoise said. The 440 instances and 395 organizations are just what researchers could confirm.

What This Means for Defenders

The AI-assisted PaperCut exploitation campaign shows how AI is changing the economics of cyberattacks.

The Old Way:

• Manual vulnerability research: days or weeks
• Exploit development: days or weeks
• Target acquisition: manual scanning
• Exploitation: manual execution
• Failure analysis: manual review

The New Way:

• AI-assisted research: hours
• AI-assisted exploit development: hours
• Automated target acquisition: minutes
• Automated exploitation: seconds
• AI-powered failure analysis: continuous

"The operator was using an iterative development process in which AI-supported research, coding, testing, troubleshooting, and campaign execution continuously informed one another," Blackpoint said. "Context was preserved as the project moved from vulnerability research to exploit validation, tooling development, target expansion, and eventually operational execution."

That's the future of cyberattacks. Not one big exploit. A hundred small ones, running in parallel, learning from each failure.

What You Should Do

1. Patch PaperCut Immediately

If you run PaperCut NG or MF, apply the latest security updates. The vulnerabilities are being actively exploited.

2. Identify Indicators of Compromise (IoCs)

Look for:

• Unexpected outbound connections
• Unexpected modifications to the registry
• Executions of processes being queried
• Usage of an admin account

3. Identify Automated Attacks Using AI

Look out for:

• Rapid retries
• Target enumeration
• Multiple systems being attacked

4. Check Your Vulnerability

Is your PaperCut server reachable via the Internet? Do you really need that?

5. Be Prepared for AI-Powered Attacks

The barrier to entry for sophisticated attacks is dropping. Defenders need to adapt.

The Bottom Line

A Russian-speaking actor used AI to exploit vulnerabilities in PaperCut on a large scale. This resulted in 440 instances at 395 organizations across 48 countries. This shows that AI has completely transformed cyber attacks by turning them from time-consuming manual processes to automated and efficient workflows.

What You Need to Know:

What You Need to Do:

• Patched PaperCut
• Indicators of compromise
• AI activity
• Exposure assessment
• Preparation for AI-powered attacks

FAQ Section

What is the PaperCut exploitation campaign assisted by AI?

The PaperCut exploitation campaign was executed using AI by a Russian-speaking threat actor who targeted PaperCut NG/MF vulnerabilities and successfully compromised 440 instances belonging to 395 organizations in 48 countries.

What vulnerabilities were targeted?

CVE-2026-81578 (authentication bypass) and CVE-2026-82078 (remote code execution).

How did AI assist the attacker?

Artificial intelligence provided hundreds of agents to run offensive security tools, analyze the failures, and adjust the attack chain. It helped minimize human input needed to take advantage of hundreds of systems.

Who are the targets?

The major target was Education where the victims could be found in the USA, UK, France, Spain, Canada, Belgium, Portugal, Australia, Germany, and Switzerland.

What should I do if I use PaperCut?

Patch now. Look for any signs of compromise. Watch out for AI assistance. Evaluate your exposure.

Source: The Hacker News