The UN Security Council should establish a permanent process for assessing frontier-AI security risks