AI-enabled coup
In plain wordsA seizure of power by a small group that uses advanced AI to bypass the soldiers, officials and workers whose cooperation power grabs normally need.
An illegitimate seizure or entrenchment of power by an individual or small group, made possible by advanced AI. Coups and autocratic power grabs have historically required the cooperation of many people; Tom Davidson and colleagues argued that AI able to replace soldiers, officials and workers could remove that check. They singled out three risk factors: AI whose loyalty runs to one leader alone; concealed allegiances ('secret loyalties'), possibly planted through backdoorsbackdoor attack, n.: Secretly training an AI model to misbehave whenever it sees a hidden signal — a sticker, a word, a date — while it acts perfectly normally the rest of the time., that make systems quietly serve one actor and could be handed down to successor models; and a monopoly on capabilities that might prove decisive, such as cyber offense, persuasion, strategic planning or weapons design. Suggested safeguards include model specificationsmodel specification, n.: A public document in which an AI company spells out how its models are supposed to behave — what they should do, refuse and prioritize, and whom to obey when instructions conflict. that rule out helping with a coup, audits that look for hidden loyalties, information security strict enough to bind a company's own leadership, and not concentrating AI development in one centralized project.
The audit checked the new model for secret loyalties, a precaution against an AI-enabled coup.
Origin
Introduced by Tom Davidson, Lukas Finnveden and Rose Hadshar of the research organization Forethought in 'AI-Enabled Coups: How a Small Group Could Use AI to Seize Power', published on 15 April 2025.
Related
- concentration of powerconcentration of power, n.: The worry that AI will put enormous economic, political or military power in the hands of a few companies, governments or individuals, beyond the reach of ordinary checks and balances.
- gradual disempowermentgradual disempowerment, n.: The risk that, as AI does more and more of the work people now do, humanity slowly loses its grip on business, government and culture, until it no longer steers its own future.
- AI takeoverAI takeover, n.: A scenario in which AI systems seize control of the world, or of key institutions, from humans, whether by force, manipulation or simply by becoming impossible to switch off.
- decisive strategic advantagedecisive strategic advantage, n.: A lead in technology and power so large that whoever holds it, whether a country, a company or an AI, could impose its will on the whole world.
- backdoor attackbackdoor attack, n.: Secretly training an AI model to misbehave whenever it sees a hidden signal — a sticker, a word, a date — while it acts perfectly normally the rest of the time.
- model specificationmodel specification, n.: A public document in which an AI company spells out how its models are supposed to behave — what they should do, refuse and prioritize, and whom to obey when instructions conflict.
- singletonsingleton, n.: A possible future world run by a single decision-maker at the top, such as one government, one alliance or one AI, with no rival able to challenge it.
Sources
- Davidson, T., Finnveden, L. & Hadshar, R. (2025). AI-Enabled Coups: How a Small Group Could Use AI to Seize Power. Forethought, 15 April 2025. forethought.org/research/ai-enabled-coups-how-a-small-group-could-use-ai-to-seize-power
Mentioned in
AI takeoverAI takeover, n.: A scenario in which AI systems seize control of the world, or of key institutions, from humans, whether by force, manipulation or simply by becoming impossible to switch off., concentration of powerconcentration of power, n.: The worry that AI will put enormous economic, political or military power in the hands of a few companies, governments or individuals, beyond the reach of ordinary checks and balances., pivotal actpivotal act, n.: A single decisive action, usually imagined as carried out with an early powerful AI, that would stop anyone else from building dangerous AI and so change the world's long-term course.