A smart city aspires to enhance quality of life, optimize city operations, and promote economic growth with the use of AI/ML techniques. However, the AI/ML techniques themselves often produce carbon emission due to their high demand for computations during their training. Environmentally sustainable Smart Cities require systematic measure of its carbon footprint and approaches to reduce carbon emission from cities backbone edge networks and cloud data centers. This work studies the methods and challenges in gauging the carbon footprint produced by the AI/ML implementations in smart cities.