This paper delves into how artificial intelligence (AI) facilitates authoritarianism across various political regimes, a phenomenon not limited to autocracies. By investigating the lifecycles of six AI systems deployed in diverse political contexts, ranging from the US to China, we provide enhanced transparency.
Utilizing a wide array of sources (academic publications, investigative reports, third-party evaluations, media interviews, and government procurement notices), we conduct a systematic qualitative comparison to identify the critical technical and operational features that enable authoritarianism within their respective political contexts. Key findings include:
- Centralization and co-optation of administrative data for law enforcement and political punishment.
- Regulatory gaps that fail to deter misuse.
- Weak user compliance that nullifies human oversight mechanisms.
- Encoding of protected group traits identifying members of vulnerable populations.
Our findings reveal that these features are present across systems in both autocratic and democratic regimes, albeit in varying configurations. Furthermore, both centralized and fragmented AI systems can contribute to authoritarianism by exploiting governance gaps: centralized systems directed by executive authorities, particularly within security and military institutions, often lack formal oversight mechanisms, while fragmented systems diffuse accountability among stakeholders, paving the way for entrenchment. We conclude with recommendations for developers and policymakers to mitigate these risks.
Blogger's Review: This paper provides an in-depth examination of the potential authoritarian features of AI systems across different political regimes, highlighting the profound impact of design and operational choices on political environments. Developers and policymakers must take these risks seriously and implement effective measures to ensure the sustainable and responsible development of technology.