The implementation of Internet-of-Things (IoT) in agriculture unlock the next-level of productivity in a sustainable system. In a nutshell, it brings along a myriad of cybersecurity and privacy concerns associated with the data and systems of the agricultural networks, the solutions to which must be found in order for the sector to be secure and stable. This study scrutinizes possible dangers of cybersecurity and data sharing in agricultural IoT systems and offers solutions on how to improve security and data-sharing policies. Tests with diverse, realistic agricultural IoT datasets have enabled us to examine the efficacy of the machine learning techniques and the access control approaches as well as the data encryption techniques in terms of keeping the farming data from being gambled with. Results demonstrate high accuracy in intrusion detection (95%), robust performance in anomaly detection (precision: Powerful authentication (successful login rate: 99%, intrusion detection rate: 2%) including effective access control enforcement (successful access - 98%, unauthorized access - 2%) and compliant data encryption overhead implementation (5 ms encryption time, 7 ms decryption time) resulted in reducing the number of security issues.