<p>Additive manufacturing (AM) is an accessible and sustainable process that rivals conventional manufacturing techniques. The unique design flexibility and wide range of material feedstocks make it a transformative technology, but its success depends on effective process control, design strategy, and material selection. Although AM is rapidly advancing, there remains a need to assess the environmental impacts of this technology, particularly with respect to direct emissions from the printing process and the use of flexible materials. This study proposes a data-driven framework to predict emissions from the fused deposition modeling printing process through integration of machine learning (ML) and sensor-based monitoring. The framework consists of data input, analytical, and decision-making layers that collectively enable the collection of real-time data, prediction of emissions, and optimization of process parameters. A case study using polylactic acid (PLA) demonstrates the framework’s application, where low-cost sensors capture carbon dioxide (CO₂) and volatile organic compound (VOC) emissions; multiple ML models were evaluated for prediction accuracy. Results reveal that Gaussian process regression achieved the highest performance for CO₂ prediction, while Support Vector Machines outperformed other models for VOC estimation. Extruder temperature, print speed, and infill density were identified as critical factors for emissions, thus paving the way for opportunities for process optimization. Future work will extend this framework to flexible materials, such as flexible PLA and thermoplastic polyurethane, to address the existing gap in environmental characterization of soft polymers. Overall, the proposed framework provides a foundational step toward AI-enabled, real-time environmental assessment for sustainable additive manufacturing.</p> Graphical Abstract <p></p>