Fossil fuel CO2 emissions (ffCO2) constitute the majority of greenhouse gas emissions and are the main determinant of global climate change. The COVID-19 pandemic caused wide-scale disruption to human activity and provided an opportunity to evaluate our capability to detect ffCO2 emission reductions. Quantifying changes in ffCO2 levels is especially challenging in cities, where climate mitigation policies are being implemented but local emissions lead to spatially and temporally complex atmospheric mixing ratios. Here, we used direct observations of on-road CO2 mole fractions with analyses of the radiocarbon (14C) content of annual grasses collected by community scientists in Los Angeles and California, USA to assess reductions in ffCO2 emissions during the first two years of the COVID-19 pandemic. With COVID-19 mobility restrictions in place in 2020, we observed a significant reduction in ffCO2 levels across California, especially in urban centers. In Los Angeles, CO2 enhancements on freeways were 60 ± 16% lower and ffCO2 levels were 43-55% lower than in pre-pandemic years. By 2021, California’s ffCO2 levels rebounded to pre-pandemic levels, albeit with substantial spatial heterogeneity related to local and regional pandemic measures. Taken together, our results indicate that a reduction in traffic emissions by ~60% (or 10-24% of Los Angeles’ total ffCO2 emissions) can be robustly detected by plant 14C analysis and pave the way for mobile- and plant-based monitoring of ffCO2 in cities without CO2 monitoring infrastructure such as those in the Global South.