Comment by lf88

Well..clearly it isn't safer: alignment is ill defined, and open weights models can be abliterated with limited resources.