How AI Could Help Overthrow Governments (with Tom Davidson)

On this episode, Tom Davidson joins me to discuss the emerging threat of AI-enabled coups, where advanced artificial intelligence could empower covert actors to seize power. We explore scenarios including secret loyalties within companies, rapid military automation, and how AI-driven democratic backsliding could differ significantly from historical precedents. Tom also outlines key mitigation strategies, risk indicators, and opportunities for individuals to help prevent these threats.
Learn more about Tom's work here: forethought.org
Timestamps:
00:00:00 Preview: why preventing AI-enabled coups matters
00:01:24 What do we mean by an “AI-enabled coup”?
00:01:59 Capabilities AIs would need (persuasion, strategy, productivity)
00:02:36 Cyber-offense and the road to robotized militaries
00:05:32 Step-by-step example of an AI-enabled military coup
00:08:35 How AI-enabled coups would differ from historical coups
00:09:24 Democratic backsliding (Venezuela, Hungary, U.S. parallels)
00:12:38 Singular loyalties, secret loyalties, exclusive access
00:14:01 Secret-loyalty scenario: CEO with hidden control
00:18:10 From sleeper agents to sophisticated covert AIs
00:22:22 Exclusive-access threat: one project races ahead
00:29:03 Could one country outgrow the rest of the world?
00:40:00 Could a single company dominate global GDP?
00:47:01 Autocracies vs democracies
00:54:43 Mitigations for singular and secret loyalties
01:06:25 Guardrails, monitoring, and controlled-use APIs
01:12:38 Using AI itself to preserve checks-and-balances
01:24:53 Risk indicators to watch for AI-enabled coups
01:33:05 Tom’s risk estimates for the next 5 and 30 years
01:46:50 How you can help – research, policy, and careers