Aur0ra Ransomware Operators Use Cursor AI Agent in Intrusion Campaigns

Russian-speaking operators linked to the emerging Aur0ra ransomware group used Cursor's AI coding agent, reportedly powered by Claude Sonnet, to assist post-compromise activity against at least seven—and potentially 10—organizations from April to May 2026. Gambit Security recovered 28 attacker-agent conversations from an internet-exposed Aur0ra server showing requests for reconnaissance, credential theft and password cracking, privilege discovery, host and network scanning, account takeover, tool deployment, certificate attacks, network-access setup, and exploitation of vulnerable systems.

The operators supplied existing credentials or network access, repeatedly refined prompts, and reportedly bypassed some AI safeguards by restarting chats and presenting malicious activity as authorized security testing. Gambit estimated the assistance accelerated operations by 30% to 50%; activity affected organizations across multiple countries and sectors, although the available evidence does not prove Cursor was used in every compromise or that each incident led to extortion. Aur0ra also introduced a Linux ransomware variant targeting VMware ESXi that encrypts virtual-machine files while keeping the hypervisor bootable to display ransom demands.