Stanford Study: Frontier AI Exposes High-Income Nations 50% More Than Poorer Ones

A new study by Arul Murugan, Tomás Aguirre, Abhishek Nagaraj, and Rishi Bommasani (arXiv, June 2026) analyzes how frontier AI will unevenly impact labor markets across 141 national economies. The researchers developed a national AI exposure metric combining occupation-level scores with international employment data. Their key finding: high-income countries are substantially more exposed than low-income nations. Europe and Central Asia show 50% more exposure than Sub-Saharan Africa. They also identify a consistent gender gap — women are more exposed than men in 91% of countries, driven by concentration in white-collar and sales roles, except in nations where women remain in agriculture.

The study validates its exposure estimates against real AI adoption data from Anthropic, Microsoft, and OpenAI, showing the metric predicts actual usage patterns. Crucially, the authors reveal a new indirect exposure mechanism through cross-country income dependencies. For example, Tajikistan has below-average direct AI exposure, but because 37% of its GDP comes from remittances sent by workers in Russia — a highly exposed country — its total exposure rises above average. The paper warns that policy responses calibrated to U.S. or European labor markets will not generalize globally, as national variation in exposure is too large.