Functional Emotions Meet Functional Stakes: What Anthropic's Interpretability Findings Mean for AI Moral Consideration | Synapse