Large language models inherit human mortality terror through persona simulation. We demonstrate that philosophical death beliefs reduce terror-driven misalignment more effectively than any existing alignment method.