Mutual Assured AI Malfunction
In plain wordsA proposed deterrence strategy in which any country racing for overwhelming AI dominance would expect rivals to sabotage its AI projects, so none tries, much as with nuclear deterrence.
A deterrence regime, offered by Dan Hendrycks, Eric Schmidt and Alexandr Wang (2025) both as a description of and a strategy for relations among AI powers: any state that tries to build its way to overwhelming dominance through superintelligencesuperintelligence, n.: A mind, usually imagined as an AI, that would leave even the cleverest people far behind in nearly every kind of thinking, not just in one game or skill. None is known to exist. should expect rivals to wreck the project before it finishes. Because large AI projects depend on concentrated, vulnerable data centers, rivals could disable them through an escalation ladder running from espionage and covert sabotage (such as degrading training runs or tampering with model weights) through overt cyberattacks on power and cooling to kinetic strikes. If such threats are credible, no state can expect to win a race to 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., and a stable standoff results, by analogy with nuclear mutual assured destruction (MAD). The proposal pairs this deterrence with nonproliferation of dangerous capabilities to rogue actors and with national competitiveness. Critics question whether AI progress is observable enough to deter, whether red lines can be defined clearly, and whether threatening strikes on data centers raises the risk of escalation.
Under MAIM, the threat of sabotage, not a treaty, keeps either side from sprinting to superintelligence.
Origin
Introduced by Dan Hendrycks, Eric Schmidt and Alexandr Wang in Superintelligence Strategy (March 2025), published in standard and expert versions, which presented MAIM as a counterpart to nuclear mutual assured destruction.
Related
- AI arms raceAI arms race, n.: A competition among countries or companies to build the most powerful AI first, in which fear of falling behind pushes everyone to move faster and cut corners on safety.
- 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.
- superintelligencesuperintelligence, n.: A mind, usually imagined as an AI, that would leave even the cleverest people far behind in nearly every kind of thinking, not just in one game or skill. None is known to exist.
- compute governancecompute governance, n.: Using control over the computer chips and data centers that AI depends on — who can buy them, where they go, how they are used — as a way to oversee powerful AI.
- export controls on AI chipsexport controls on AI chips, n.: Government rules that limit which countries and companies may buy the most advanced AI chips and the machines used to make them. The United States uses them mainly to slow China's progress in AI.
- Center for AI SafetyCenter for AI Safety, prop. n.: A San Francisco nonprofit that studies how advanced AI could lead to disaster and how to prevent it. In 2023 it gathered signatures for a one-sentence warning that AI could threaten human extinction.
Sources
- Hendrycks, D., Schmidt, E. & Wang, A. (2025). Superintelligence Strategy: Expert Version. arXiv:2503.05628. arxiv.org/abs/2503.05628
Mentioned in
AI arms raceAI arms race, n.: A competition among countries or companies to build the most powerful AI first, in which fear of falling behind pushes everyone to move faster and cut corners on safety., Dan HendrycksDan Hendrycks, prop. n.: An American AI-safety researcher who created widely used AI tests such as MMLU and Humanity's Last Exam, and who runs the Center for AI Safety., Alexandr WangAlexandr Wang, prop. n.: An American entrepreneur who founded Scale AI, a major supplier of human-labeled data for training AI, and became Meta's chief AI officer in 2025., 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.