I never argued for restrictions (I don't agree with the article authors) and it's a real risk if powerful AI is only controlled by already powerful entities. My comment was about the dangers of equating AI safety with aligning chatbots and censoring diffusion models.
