mattmacdermott comments on On OpenAI’s Safety and Alignment Philosophy