The race for effective AI war machines was always inevitable, but portends a horrible future. I don't see how to avoid it. When states see their adversaries r

The race for effective AI war machines was always inevitable, but portends a horrible future. I don't see how to avoid it. When states see their adversaries researching murderbots, how can they decline to maintain a balance of power by doing the same? In the context of competing states, how can the ethical discussion about murderbots have any influence?