An AI researcher’s departure over concerns about increasingly powerful, self-improving systems is reigniting the debate over AI safety. Chris insists that even a small risk of catastrophic consequences warrants serious guardrails—and warns against racing ahead with technology before understanding what could go wrong.