Global AI Governance Timeliness — LessWrong