Food transportation is a critical yet under-optimized component of United States food security, linking farms, processors, distribution centres, retailers, food banks, and consumers across geographically dispersed and disruption-prone networks. Existing systems are constrained by fragmented data, empty backhauls, cold-chain failures, port and highway congestion, driver shortages, severe weather, fuel-price volatility, and unequal access in rural, low-income, and disaster-affected communities. This study develops an artificial intelligence-driven optimization framework tailored to US food logistics. The framework integrates machine-learning demand forecasts, real-time traffic and weather data, vehicle-routing algorithms, predictive maintenance, dynamic fleet allocation, and IoT-based temperature monitoring to coordinate refrigerated and non-refrigerated food movements. It prioritizes perishability, nutritional importance, delivery urgency, infrastructure capacity, and regional vulnerability when generating transport decisions. The proposed system also models disruption scenarios, including hurricanes, wildfires, floods, cyber incidents, and distribution-centre outages, enabling rapid rerouting and inventory rebalancing across states. Expected outcomes include lower spoilage, fewer stockouts, reduced vehicle miles, improved cold-chain compliance, faster emergency food delivery, and more reliable service to underserved areas. By embedding resilience, equity, and sustainability into transport optimization, the framework provides a practical pathway for strengthening national food availability, reducing logistics costs, and improving the overall responsiveness of the US food supply chain.