<p>In the current global digital wave, the traditional agricultural basic service model is undergoing changes, and the integration of artificial intelligence and Internet of Things technology has brought new opportunities for agricultural modernization. This article aims to explore how to use artificial intelligence and IoT technology to build a network security protection system, in order to promote the sustainable economic development of agricultural basic services. A collaborative intrusion detection model based on contrastive federated learning (ID-CFL) was proposed in the study. This model uses federated learning to train IoT devices collaboratively without sharing raw data, and introduces contrastive learning to enhance the recognition ability of potential intrusion behaviors, especially for detecting unknown abnormal patterns. The experimental results show that the ID-CFL model exhibits superior accuracy on different datasets, effectively improving the performance of IoT intrusion detection. The study used methods such as SBM-GML index to evaluate agricultural green total factor productivity and found that the collaborative application of artificial intelligence and IoT technology significantly improved agricultural production efficiency and decision-making accuracy. In addition, the development of the digital economy has had a positive impact on the agricultural sector, promoting sustainable development through optimizing resource allocation, improving technological efficiency, and fostering technological innovation. Therefore, building a safe and reliable intelligent agricultural service system is crucial for addressing the cybersecurity challenges in the digital transformation of agriculture. This will help promote the development of agricultural production towards efficiency, environmental protection, and sustainability, laying the foundation for long-term stable growth of the agricultural economy and global food security.</p>