AI Agents Used to Compromise 440+ PaperCut Instances — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- PaperCut NG/MF instances were compromised at 395 organizations across 48 countries using hundreds of AI agents powered by OpenAI Codex and a DeepSeek model, exploiting CVE-2026-81578 and CVE-2026-82078 as an auth-bypass-to-RCE chain
- GreyNoise and Blackpoint Cyber traced the activity to IP address 45.142.193[.]132, tracking the same address since early July 2026 for probing systems from Palo Alto, Ubiquiti, Citrix, SonicWall, and Proxmox VE vendors
- The actor moved from an empty workspace to remote code execution against a real victim in under four hours, then compromised 11 organizations in 26 seconds once the campaign launched; only 12 of the 395 organizations reached full domain admin access
- In one U.S. high school attack, initial access escalated to full domain administrator access in seven minutes, while post-exploitation tools included Metasploit/Meterpreter Java payloads, Windows registry hive collectors, and Mimikatz, SharpHound, Certipy, Rubeus, and Impacket
- The actor built a self-hosted lab with vulnerable PaperCut software and an Active Directory server for exploit development, used Netlas.io via an identified API key to build target lists, and explicitly attempted to exclude 28 countries including Russia, China, Iran, and Venezuela
- Two open-source tools powered the orchestration: Hindsight provided persistent memory across AI agents, while AionUi offered a unified workspace to run and view multiple agents concurrently through research, exploit development, target filtering, and execution stages
Why it matters: This campaign demonstrates AI is no longer just assisting attackers but orchestrating entire exploitation pipelines — from vulnerability research to target filtering to post-exploitation — compressing timelines to as little as seven minutes from initial access to domain admin. By reducing human effort to oversight across 395 targets, the economics of cyberattacks shift: smaller crews can now run campaigns at a scale and speed that previously required larger teams, reshaping the defensive calculus for the education sector that bore the brunt of these attacks.
Ask SkimNews
