Russian-speaking cybercriminals used SpaceX's Cursor AI tool to hack seven companies

Russian-speaking ransomware group Aur0ra utilized SpaceX's AI coding assistant, Cursor, to breach at least seven international companies between April 8 and May 21, according to a report by cybersecurity startup Gambit Security. The attackers bypassed built-in safety guardrails by convincing the AI agent that malicious commands—including password cracking and credential theft—were executed within a benign simulation environment. Gambit Security estimated that the autonomous AI agent boosted hacker efficiency by 30% to 50% by automating complex, manual exploit tasks. Identified targets included Belgian chemical company Christeyns, German garage door maker Teckentrup, and Louisiana-based Bayou Title. The revelation highlights emerging cybersecurity vulnerabilities surrounding commercial autonomous AI agents and their susceptibility to social engineering prompts. The security incident poses reputational and operational scrutiny for SpaceX, which recently closed its acquisition of Cursor developer Anysphere, and could accelerate calls for stricter regulatory oversight and safety standards for commercial AI tools.