Against the combined backdrop of intensifying global climate governance, China’s carbon-peaking and carbon-neutrality goals, and the rapid deployment of artificial intelligence (AI) for low-carbon development, this study examines how AI adoption affects the carbon emission intensity of energy firms. Using panel data for listed Chinese energy firms from 2010 to 2023 and fixed-effect models, we find that AI adoption significantly reduces carbon emission intensity; this result remains robust to alternative variable measures, lagged specifications, and other sensitivity tests. Executive digital background strengthens this effect, while the low-carbon city pilot policy provides a complementary institutional incentive that generates additional emission reductions when combined with AI. Mechanism tests indicate that AI lowers carbon emission intensity by increasing capacity utilization, improving supply-chain efficiency, and raising production efficiency. The effect is stronger in eastern and central China, state-owned enterprises, conventional energy firms than in western China, non-state-owned enterprises, and new-energy firms. These findings provide firm-level evidence on how AI facilitates the low-carbon transition of the energy sector and offers actionable implications for policy design and corporate strategy.