<p>Understanding the wettability of carbon dioxide (CO₂) and the interfacial properties between reservoir rocks and fluids is crucial for the effective geological carbon sequestration (GCS). The most accurate way to measure these properties is the laboratory experiments under simulated reservoir conditions. However, experimental measurement of CO₂ wettability in storage/caprock, influenced by thermo-physical conditions, poses significant challenges due to the reactivity and embrittlement risks associated with high levels of CO₂. Therefore, data-driven machine learning (ML) models can be used as an alternative to laboratory experiments to predict rock/CO₂/brine wettability in a precise and less hazardous manner. In this study, we have used multiple ML models, including stacked generalization regression (SGR), gradient boosting, and tree-based models, to estimate the wettability of Saudi Arabian (SA) basalt within a ternary system involving rocks, CO₂, and brine, operating under diverse conditions. To improve the accuracy of the ML models, a comprehensive set of experimental data was collected from the literature that covered a wide range of pressure and temperature, 0.1–25&#xa0;MPa and 298–343&#xa0;K, respectively. Various data exploration methods, such as heatmaps, and histograms were used to thoroughly examine the laboratory dataset. The ML models were trained to predict the advancing and receding contact angles. The results showed that the proposed ML models could accurately forecast wettability behaviors across diverse operational conditions with an average R² score of above 0.996. The outcomes of ML models can be highly useful for accurately determining the CO₂ wettability. This information is crucial for defining the storage capacity and assessing containment security in large-scale CO₂ sequestration projects.</p>