Metal-organic frameworks (MOFs) have attracted significant attention for their versatile applications, particularly in gas adsorption and separation. Developing accurate force fields is crucial for reliably predicting the behavior of various phenomena in MOFs. We present a novel approach for generating interatomic force fields tailored for UiO-66, a zirconium-based MOF, utilizing advanced machine-learning methods, namely Machine Learning Interatomic Potentials (MLIP) and the Multi-Atomic Cluster Expansion (MACE). Through an active learning framework, we iteratively refine the force fields to ensure high accuracy, robustness, and computational efficiency, achieving near DFT calculation precision. The developed MLIP- and MACE-based force fields successfully enable detailed molecular dynamics simulations, accurately capturing structural, mechanical, and adsorption properties of UiO-66. The methodology demonstrates significant improvements in predictive power, providing an effective tool for exploring the properties and design of MOFs.