The pandemic of COVID-19 and the global policy implemented to attenuate the spread of the virus, provided a unique opportunity to study the footprint of the resulting social impact of COVID-19 pandemic on seismic data. In this study, a non-extensive statistical physics approach is examined for anthropogenic high-frequency seismic noise in view of COVID-19 pandemic lockdown measures and the consequent changes in human activity levels. A continuous monitoring of the entropic index q performed during a time period which includes a time interval of normal social activity and a period of anthropogenic quiescence due to the COVID-19 pandemic lockdown measures, enabled us to observe a dependance between the human activity levels and the degree of non-extensivity in the seismic signal. Average values of pre-lockdown q parameters estimated within the urban environment of the cities of Athens and Thessaloniki (Greece), were equal to 1.37 and 1.56, respectively, while the respective post-lockdown q values were found to be 1.25 and 1.49. In overall, the estimated q parameters could support the conclusion that the anthropogenic seismic noise is sub-extensive. Lower values of q during the lockdown period reflect a lower degree of interacting subsystems and degree of correlations in the high-frequency seismic noise wavefield. A short-term examination on a week scale of the temporal behavior of parameter q, revealed significant patterns of daily periodicity and diurnal variations resembling the diurnal cycle of human activities. According to the preliminary observations of this study, the non-extensive parameter q is shown to be a rather promising tool regarding the quantitative strategies for monitoring the effect of the anthropogenic activity levels in seismic data.