The rising levels of CO2 emissions are a pressing concern for nations globally, necessitating advanced solutions for monitoring and reduction. Biogenic feedstocks coprocessing is a cost-effective method for oil refineries to produce lower carbon intensity fuels and reduce the CO2 emissions. In order to quantify the environmental impact of the oil refining process, it is crucial to monitor the green CO2 emissions that produced by biogenic feedstocks throughout the entire co-processing process. In this study, artificial intelligence (AI) is first being adopted in the field of biogenic feedstocks coprocessing for modeling green CO2 emissions. We collected 102,000 samples, which, to the best of our knowledge, is the largest dataset of its kind. A novel framework based on machine learning is introduced to establish a data-driven model for real-time green CO2 emission tracking. To evaluate the accuracy and feasibility of various machine learning models, we employed 10 different feature analysis methods and 5 different regression methods for CO2 emission modeling. This analysis enables policymakers and stakeholders to assess the performance and environmental impact of biogenic feedstocks co-processing in renewable energy production, offering novel insights into the potential opportunities and benefits of using biogenic feedstocks co-processing.