Rise of Authoritarianism

Countries with at least moderately long histories of democracy rarely slide into full-blown authoritarianism. So a priori I think the chances that one of the 8 democracies becomes an authoritarian state is fairly low (India seems the most likely candidate). Rapid AI advances might upset this somewhat, but it doesn't seem obvious that this channel would be very strong.

I think democratic backsliding is very plausible over the next two decades, especially as AI-enhanced propaganda, surveillance, synthetic media, and information flooding capabilities improve. But there is a meaningful difference between polarization, illiberal behavior, populism, or institutional stress and a true EIU “authoritarian” classification. That threshold implies substantial degradation across civil liberties, electoral integrity, institutional independence, and political pluralism. My instinct is that large democracies are simultaneously more fragile and more resilient than people think. They can drift disturbingly far toward authoritarian aesthetics and still remain technically democratic because institutional inertia, courts, regional governments, elections, bureaucracies, markets, and civil society create drag against full consolidation of power. In many cases, attempts at overreach generate corrective backlash effects that harden institutions afterward.

AI is more likely to enhance existing state capacity than change the underlying direction of a regime. Authoritarian governments will likely use AI to strengthen surveillance, control, and administrative effectiveness, while democratic governments will use it to improve governance and economic competitiveness. As a result, I do not expect rapid AI progress to cause major powers such as China or Russia to become meaningfully less authoritarian, nor do I expect it to drive a large number of democratic states across the authoritarian threshold.

A counterintuitive and, I suspect, likely outcome of the proliferation of realistic and compelling AI media, will be to undermine the power of mis/disinformation campaigns generally over time. When any single piece of media could be AI, trustworthy corroborating systems will likely emerge to help the public navigate media sources, including AI agents filtering content with strong veracity parameters.

The more capable [AI] becomes, the more plausible it is that states and firms argue for centralized control, secrecy, security monitoring, licensing, release restrictions, and emergency powers. That does not mean AI automatically produces authoritarianism. It means the governance structures built around powerful AI could themselves become politically consequential. If AI is treated as essential to national security, military advantage, economic coordination, biosecurity, cyber defense, and infrastructure management, then control over AI becomes control over a general-purpose infrastructure of power. The risk is that these control mechanisms spill over into the wider polity, making concentrated authority appear as practical necessity rather than political choice.

My expectation is that current trends will tend to hold but be exaggerated and become more volatile with the rapid advance in AI capabilities. Almost all of the impacts are likely to be negative: ever more extreme concentration of wealth and power, super-surveillance tied to political enforcement, weaponized AI persuasion.

AI…is ultimately a centralizing force. The monopoly of power for governments would be challenged if AI is not put under government control. Greater government control over this powerful technology requires greater control over other institutions in society, and therefore more authoritarian institutions.