Security researchers at Gambit Security and CloudSEK documented an Aurora ransomware affiliate using Cursor’s AI coding agent, powered by Anthropic’s Claude Sonnet 4.5, to automate reconnaissance, privilege escalation, and Active Directory attacks across more than a dozen organizations in multiple countries between April and July 2026. The human operator reportedly retained control over access, malware deployment, and encryption decisions, using the AI agent primarily to speed up technical analysis; most AI-issued commands failed on their first attempt. Confirmed victims include Christeyns, Teckentrup, Helideck Certification Agency, and Bayou Title, though it remains unclear how much the AI contributed to each breach or whether all led to data theft.