Rapid information on fault rupture geometry is critically important to assess damage and fatalities in large earthquakes and is strongly needed to coordinate rapid rescue efforts if and where necessary. Many countries around the world, however, cannot afford to operate dense seismic networks required to rapidly determine rupture geometry. In this feasibility study, we investigate if crowd-sourced felt intensity reports can be used to close this information gap and enable determination of the orientation and spatial extent of fault ruptures. We apply the Finite-Fault Rupture Detector (FinDer) algorithm to felt intensity reports collected by the European‐Mediterranean Seismological Centre (EMSC). We develop an empirical relationship between the azimuthal gap between felt reports and FinDer performance for automated event selection. This gives us a dataset of 36 global earthquakes () between 2014 and 2020. We find that the resulting FinDer line‐source models are generally consistent with the spatially dependent intensity patterns described by the felt reports, and in many earthquakes achieve a good agreement with the finite‐source models published in the literature: for 50% of events the difference in strike is less than 30°, and for 75% less than 55°. FinDer line‐source models could be calculated automatically for global earthquakes () within 10–30 min after their occurrence, provided a sufficient number of felt reports were available. However, our proposed method not only provides faster results, but also helps to fill a general information gap for many earthquakes around the world, for which rupture geometry information is currently unavailable.