AI-Enabled Coups: How a Small Group Could Use AI to Seize Power

By Tom Davidson, Lukas Finnveden and Rose Hadshar. 

The development of AI that is more broadly capable than humans will create a new and serious threat: AI-enabled coups. An AI-enabled coup could be staged by a very small group, or just a single person, and could occur even in established democracies. Sufficiently advanced AI will introduce three novel dynamics that significantly increase coup risk. Firstly, military and government leaders could fully replace human personnel with AI systems that are singularly loyal to them, eliminating the need to gain human supporters for a coup. Secondly, leaders of AI projects could deliberately build AI systems that are secretly loyal to them, for example fully autonomous military robots that pass security tests but later execute a coup when deployed in military settings. Thirdly, senior officials within AI projects or the government could gain exclusive access to superhuman capabilities in weapons development, strategic planning, persuasion, and cyber offence, and use these to increase their power until they can stage a coup. To address these risks, AI projects should design and enforce rules against AI misuse, audit systems for secret loyalties, and share frontier AI systems with multiple stakeholders. Governments should establish principles for government use of advanced AI, increase oversight of frontier AI projects, and procure AI for critical systems from multiple independent providers.

Source: 

https://www.forethought.org/research/ai-enabled-coups-how-a-small-group-could-use-ai-to-seize-power

A podcast by BlueDot Impact.

Learn more on the AI Safety Fundamentals website.

Chapters

• 0:23 Abstract
• 1:54 Summary
• 2:45 An AI workforce could be made singularly loyal to institutional leaders
• 4:18 AI could have hard-to-detect secret loyalties
• 6:12 A few people could gain exclusive access to coup-enabling AI capabilities
• 9:26 Mitigations
• 12:39 1 Introduction
• 20:23 2 AI will have powerful coup-enabling capabilities
• 29:18 3 Risk factors of AI-enabled coup risk
• 30:28 3.1 Singular AI loyalties to institutional leaders
• 37:44 3.2 Secret AI loyalties
• 43:11 3.3 Exclusive access to coup-enabling capabilities
• 44:08 Concentration between AI projects
• 52:15 Concentration within leading AI projects
• 56:21 3.4 Interactions between drivers
• 1:01:03 4 Concrete paths to an AI-enabled coup
• 1:03:45 4.1 Coups using military AI
• 1:19:17 4.2 Conventional coups and backsliding
• 1:26:45 5 Mitigations
• 1:32:23 5.1 Establishing rules for legitimate use of AI
• 1:39:00 5.2 Technical measures to enforce these rules
• 1:50:21 5.3 Empowering multiple actors to prevent AI-enabled coups
• 1:51:12 Increasing transparency
• 1:55:41 Distributing decision-making authority
• 1:57:52 Sharing access to AI capabilities
• 2:01:39 Coup-proof plans for a centralised AI project
• 2:06:58 6 Conclusion