The pandemic lockdown resulted in billions of individuals be- coming socially isolated while being bombarded with unset- tling news through online media outlets. Emotional contagion, which can propagate through means beyond in-person contact, is likely to have played a significant role in amplifying the spread of negative emotions worldwide. Despite the fact that facial be- haviors can become more synchronous under the influence of emotional contagion, it remains unclear whether individuals ex- hibit synchronous negative facial expressions in stressful situa- tions where social contact is limited. To address this gap, this study took a naturalistic viewing approach to record real-time facial behaviors while participants watched a series of pandemic and death-related movie clips in a lockdown group (N=26) and a post-lockdown group (N=32). The facial action unit (AU) fea- tures were extracted from facial videos, and intersubject corre- lation (ISC) was utilized to compute the degree of synchrony across individuals within their respective groups. Lastly, we built a machine-learning model that utilized within-group AU- ISC values as features and accurately classified participants into their respective groups. Our results revealed that the weights of our predictive model closely aligned with expressions of sad- ness and fear, supporting our initial hypothesis that individuals tend to exhibit more synchronized facial behaviors, particularly those associated with negative emotions, during the lockdown period than during post-lockdown periods. Our study offers novel insights that could inform future research into the under- lying mechanisms and implications of synchronized facial be- haviors as a potential collective response to external stressors.