As we have no choice but to use AI to safeguard AI,redundancy and fail-safe mechanisms are required. Above all, international cooperation will restrain unintended consequences and ensure that superintelligence is aligned with human goals.