“Emil Michael believes Anthropic's restrictive policies are driven by a 'doomer' view on AI risk and values rooted in effective altruism.”