A new paper published on arXiv explores how AI systems can be designed to enable authoritarianism, even in democratic regimes. The research analyzes six AI systems across the US and China, identifying common technical and operational features that facilitate authoritarian control. These include the misuse of administrative data, regulatory loopholes, weak user oversight, and the identification of vulnerable populations. The paper concludes that both centralized and fragmented AI systems can contribute to authoritarianism by exploiting governance gaps, and offers recommendations for developers and policymakers to mitigate these risks. AI
IMPACT Highlights how AI design choices can inadvertently or intentionally support authoritarian practices, urging policy and developer action.
RANK_REASON Research paper published on arXiv detailing findings on AI systems and authoritarianism. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →