With all the AI safety talks going on, I think one of the key points that’s being overlooked, is that many new voters will consult LLMs regarding whom to vote for. Such models can be turned into propaganda machines.