Remote oil and gas assets are increasingly operated through connected instrumentation, supervisory control platforms, historians, cloud analytics and integrated operations centres. This digital shift improves production surveillance, artificial-lift optimisation, reservoir management and pipeline integrity, but it also creates a cyber-physical attack surface in which manipulated telemetry, unauthorised commands, delayed alarms or compromised remote access can produce safety, environmental and production consequences. This review develops an artificial intelligence (AI)-based threat detection and response model for remote wellheads, electric submersible pump (ESP) systems and pipeline monitoring environments. Evidence published between 2020 and 2025 is synthesised across operational technology (OT) security, industrial anomaly detection, petroleum analytics, Permanent Downhole Monitoring Systems (PDHMS), downhole gauges, variable-frequency drive (VFD) monitoring and critical-infrastructure guidance. The review argues that PDHMS should not be treated as a separate topic; bottom-hole pressure, downhole temperature, vibration, intake and discharge pressure, motor current and real-time production data are integrated evidence sources for cyber-physical resilience. The proposed model combines data assurance, process-aware baselines, network intrusion detection, asset criticality, explainability, and operator-approved response playbooks. For wellheads, the key requirement is to detect command and pressure-flow inconsistencies before unsafe valve or choke action occurs. For ESP systems, cyber analytics must distinguish genuine pump degradation from malicious manipulation of VFD settings, vibration, motor current, frequency and downhole gauge data. For pipelines, detection must integrate SCADA anomalies, leak indicators, geospatial evidence and communications integrity. The review therefore prioritises cyber-physical consequence rather than alert volume, linking AI outputs to verification, containment, safe-state operation, recovery and continuous learning.