ERA5 t2m issue?

I downloaded two times in April and May netcdf data for ERA5 monthly t2m 1940-2023. The average global means seems to be way lower ranging from ~277 K to 279K over the yearly average time series, or 274.94 K 281.79 K for the monthly time series (without monthly averaging per year).

I tried redownloading the data, used Matlab and R and Panoply and different libraries to check for potential differences in reading the data, but I still get this low values. I also the same result with GRIBB datasets.

Is there an issue with the data ? or some offset/scale factor that I am not fully realizing. This is way lower than the published trends of global temperature.