Artificial Intelligence Empowers Autocrats While Empowering a New Technical Elite
Authoritarian regimes leverage machine learning to automate governance but risk dependency on technologists.
1Automation of State Propaganda
Following on-air protests on Russian state television, broadcaster RT began replacing human news anchors with artificial intelligence models in October 2024. State propagandists noted that synthetic presenters eliminate wage demands, insurance requirements, and the persistent political vulnerability of on-air dissent by human personnel.
2Resolving the Subordinate Dilemma
Autocrats historically faced a chronic tradeoff between delegating operational authority and preventing bureaucratic treason or incompetence. Machine learning models offer rulers direct oversight, automating surveillance tasks and replacing discretionary administrative decisions with centralized algorithms that theoretically cannot be bribed or convinced to leak state information.
3Automated Anti-Corruption Audits
China deployed its Zero Trust AI system across roughly 30 regional administrative districts, cross-referencing more than 150 official databases. The automated audit detected embezzlement and nepotism among 8,700 local bureaucrats, eventually forcing local reductions in deployment after bureaucratic resistance pushed back against continuous surveillance.
4Expanding Algorithmic Policing Tools
Moscow integrated facial recognition across 200,000 security cameras to detain protesters during the 2021 demonstrations. By 2023, the Internet regulator introduced the Oculus system to scan hundreds of thousands of images daily for political dissent, supplementing human censors with scalable computer vision screening.
5Algorithmic Legislative Policy Initiatives
The United Arab Emirates established its Regulatory Intelligence Office in April 2025 to draft and amend federal statutes using automated tools. By 2026, the federal leadership announced plans to automate half of government operations within two years using autonomous computational software agents.
6Opaque Digital Praetorian Guard
Algorithmic governance cannot operate autonomously because it requires ongoing optimization by specialized data scientists and software engineers. Political executives lack the technical literacy required to evaluate machine learning architectures, transferring significant gatekeeping power to a concentrated cadre of technical personnel who manage data pipelines.
7Single Points of Failure
Unlike historic standing armies or broad imperial guards, modern machine learning systems rely on a very small group of specialists. This extreme concentration creates severe systemic vulnerabilities, as disrupting automated surveillance or delaying real-time intelligence dashboards requires coordination among only a few vital system architects.
8Concessions to Retain Specialists
Following the 2022 invasion of Ukraine, Russia exempted tech workers from military conscription, eliminated corporate taxes for IT enterprises, and subsidized mortgages. Despite approximately 100,000 specialists departing, the state relied on economic incentives because specialized engineering capabilities cannot be effectively conscripted through coercion.
9Informational Sabotage in Crises
Engineers who manage intelligence dashboards hold structural leverage during domestic leadership challenges or military mutinies. By subtly delaying data ingestion, degrading live video feeds, or suppressing alerts, technical operators can blind executive authorities without openly participating in kinetic military actions or armed insurrections.
10The Illusion of Autonomy
While artificial intelligence lowers the marginal cost of public surveillance and censorship, it creates deep systemic vulnerabilities for autocracies. Rulers attempting to bypass unreliable human bureaucrats merely transfer their structural dependency onto software engineers who understand their own indispensability within the modernized security apparatus.