AI Policy Tuesday: The Case for Regulating AI Companies, Not AI Models — LessWrong