As the number of connected devices increases, protecting sensitive information transmitted by IoT devices is critical. One of the most effective ways to ensure data confidentiality while accommodating the limited resources of constrained devices is to use lightweight cryptographic algorithms. However, managing these algorithms in IoT ecosystems is a significant challenge. To overcome this problem, we propose a new IoT framework that enables self-management of IoT data confidentiality based on the use of lightweight cryptography and the Autonomic Computing paradigm. Our approach implements the MAPE-K closed control loop, which uses a decision tree algorithm to predict the most appropriate lightweight encryption algorithm based on the state of the IoT object, thus ensuring data confidentiality, improving system performance, and extending the lifetime of IoT devices.