According to CoinMeta, as reported by Forkast, 88% of IT leaders have reported at least one security vulnerability involving AI proxies in the past 12 months. Despite this frequency, the industry remains trapped in a cycle of reactive deployments. The Ghostjacking guidance recommendations released by NGCERT on October 6, 2026, aim to break this cycle by shifting the focus from detection to structural prevention. These recommendations formalize nine steps, including default denial of network access, manual approval, and the use of the least privilege model with context-aware protocols for MCP servers. The core directive is clear: stop attempting to detect malicious intent and begin to limit the operational capabilities of proxies. Survey data shows that although 86% of IT and security decision-makers regard AI proxies as critical, only 37% have established formal AI policies. 54% of organizations report having 1 to 100 unauthorized proxies, creating a complex shadow AI issue that increases the difficulty of maintaining centralized security.