As AI systems grow more advanced, we often focus on alignment, value loading, or behavioral guardrails. But what if ethics isn’t something to program in, but something that only arises structurally under specific conditions? I’ve just published a theor…