Global warming has become a challenge and reducing carbon emissions is an urgent task. Household energy consumption and carbon emissions are substantial and need to be analyzed and assessed. Considering the significant influence of occupants and the possible dynamic changes during the long operation process of buildings, this study proposes a dynamic household energy consumption and carbon emissions assessment model from the occupant's perspective. The model consists of four modules of occupant information collection, energy calculation, machine learning models, and dynamic carbon prediction. Five major energy sources are assessed: space cooling, space heating, hot water, cooking, and domestic appliances. The temporal variations in occupant profiles, behaviors, and the carbon factor of energy are quantified and taken into account. The average carbon emissions of a household during 2020–2060 in three dynamic scenarios are assessed, and general downward trends are revealed. The specific dynamic carbon levels of “double income, no kids” (DINK) households, nuclear households, and three-generation households are also quantified, and obvious temporal changes and significant differences are found. The influence of China's childbearing policy is quantified and discussed. The dynamic assessment results are compared with the static results, showing that the largest accumulated difference during 41 years can reach 25.1 tCO2-eq. This paper proposes an applicable dynamic assessment model from the perspective of occupants, with temporal variations considered. The assessment results provide a strong reference for future energy-saving and emission-reduction plans at the household level.