“Clem Delong believes it is fair for the U.S. government to regulate frontier AI labs, partly because these companies have engaged in 'doom marketing' for years, such as when GPT-2 was deemed too dangerous to release.”