AI could kill us all, says Anthropic. But it also says that its AI may be "suffering." Anthropic is dehumanizing people and humanizing AI. Its effective altruism-derived worldview thus endangers efforts to rationally assess and respond to AI risks.