LessWrong analysis: AI transition could cement executive power permanently

A recent LessWrong post by djbinder challenges elaborate AI takeover scenarios involving nanotech, drone armies, or superpersuasion. Instead, it argues that the simplest path to permanent control is through existing executive power. In the US and China, the president and general secretary already wield enormous emergency authority over military, security, and technology. During a rapid AI transition—a time of high uncertainty—they can use that authority to seize and solidify control with minimal illegal steps. Courts, Congress, and subordinates often defer to the executive in national security matters, and elections operate on longer timescales.

The post notes that naive optimization by the executive—simply trying to maintain order and prevent threats—could naturally lead to a coup-proof power structure. Once AI-enabled capabilities (e.g., surveillance, automated enforcement) are under executive control, converting them into permanent rule becomes easy. Institutional design is notoriously hard, and democracies rarely adopt reforms even when risks are clear. However, the author also points out that AI tools could empower opposition actors to monitor and resist, making the outcome uncertain. The key insight is that existing political structures, not futuristic schemes, are the most plausible vector for AI-driven authoritarian entrenchment.