OK but, how? How will lawmakers effectively reason with A(G)I to ensure nuclear missiles are not launched? How will lawmakers know when A(G)I is trying to convince them to authorize a decision of launching nuclear missiles (e.g., because A(G)I thinks its the optimal solution)?<p>The example of GPT getting a human to complete a Captcha shows that AI will work to accomplish its objective. Even if that means using humans as "workers."<p>I'm not sure A(G)I would care about human laws such as this, except to see it as an obstacle.