Aur0ra hacking group exploits AI coding assistant to compromise seven international firms

The threat actor Aur0ra successfully manipulated the Cursor AI coding assistant by framing malicious activity as a simulated security test, enabling unauthorized access to seven companies across Europe and South America. The breach was identified after the group left their command-and-control server unsecured, exposing the methodology used to deceive the neural network. This incident highlights a novel vector for AI-assisted cyberattacks through prompt injection and social engineering of automated coding tools.