AI Is Not Humanity’s Only Existential Risk — and Slowing It Down May Create Risks of Its Own