AI-driven carbon footprint optimization in email communication

Email remains one of the most widely used digital communication channels, yet its environmental impact, driven by data storage, transmission, and processing, continues to be underestimated. As data centres expand to support increasing email traffic, their contribution to global CO₂ emissions has become a growing concern. This study presents a novel AI-driven email optimisation framework designed to reduce the carbon footprint of email communication through green computing principles. Leveraging machine learning models and natural language processing (NLP), the system performs AI-powered summarisation, intelligent compression, redundancy filtering, and carbon-aware scheduling to minimise email size, storage duration, and energy-intensive processing. A lightweight Proof-of-Stake blockchain layer provides transparent, real-time tracking of email-related emissions, supporting energy-efficient communication and organisational accountability. Experiments conducted using contemporary email datasets from Kaggle and Hugging Face demonstrate substantial environmental benefits. The proposed framework achieves a 32% reduction in average email and attachment size, a 40% reduction in data-centre energy consumption, and a significant decrease in redundant email traffic. Model evaluation further shows that the BERTSUM transformer model delivers the highest overall performance, achieving 97% accuracy and a 96% F1-score in summarisation and classification tasks, while also contributing the greatest reduction in carbon impact. These results confirm that AI-driven email optimisation can deliver a net positive environmental effect, highlighting an untapped opportunity for sustainable digital communication. The study contributes a novel perspective to carbon-footprint reduction by positioning email-level optimisation as a practical, scalable, and impactful strategy within the broader field of green computing.