Offshore platforms, wind farms, and subsea pipelines are increasingly monitored and controlled through digital twins that fuse sensor data, physics-based models, and AI analytics over 5G and emerging 6G networks. These safety-critical twins demand not only ultra-reliable low-latency communication, but also robust security, privacy, and AI governance anchored in standards. This article presents a tutorial-style, standards-oriented framework for real-time monitoring and predictive control of offshore engineering assets via secure 5G/6G AI-driven digital twins. We first introduce a three-layer reference architecture spanning offshore edge devices, 5G/6G communication fabrics with URLLC and MEC, and cloud-based twin analytics, bound together by a vertical security and governance plane. Using a representative structural twin, we quantify how AI-enhanced state estimation and short-horizon forecasting enable predictive control that reduces stress exceedance risk under harsh sea conditions. We then analyze security-latency-reliability trade-offs for different protection profiles across 5G and 6G-style URLLC slices, and show how DT-aware intrusion detection and AI security can improve resilience without violating control deadlines. Finally, we map offshore DT requirements to existing standards families (3GPP, ISO/IEC 27000-series, ITU-T, industrial control security) to reveal gaps in AI governance, federated/edge learning security, and twin lifecycle management, and we outline a roadmap of best practices and standardization priorities for trustworthy DT deployments in offshore and other industrial domains.