The adoption of Internet of Things (IoT) technologies in agriculture has improved productivity through real-time monitoring, automated irrigation, and data-driven decision-making. However, increased connectivity exposes smart farming systems to cybersecurity threats that can compromise data and disrupt operations. This study proposes an AI-enabled cybersecurity framework integrating threat detection, anomaly monitoring, secure authentication, and encrypted communication. Machine learning algorithms continuously analyze device behavior and network traffic to identify cyberattacks in real time. The framework was evaluated in a simulated smart farming environment comprising sensors, edge devices, and cloud services. Results achieved 97.8% detection accuracy, 96.9% precision, and 97.3% recall, while reducing false alerts by 21% and improving response time by 28% compared with conventional methods. The proposed framework enhances security, reliability, and resilience, providing a scalable solution for next-generation digital agriculture.