We estimate the causal effect of in-person mass gatherings on the spread of Covid-19 in the United States. Using text classification on over 700,000 local-newspaper articles from 2018-2021, we construct a county-week index of reported gatherings and combine it with Covid-19 case counts, mobility, and socio-economic covariates. To address endogeneity of gatherings, we employ a shift-share instrument that interacts counties' pre-Covid propensity to generate gatherings news with the timing of Black Lives Matter protests.