US AI media attention and state-dependent spillovers across AI markets

The paper examines whether U.S. media attention to artificial intelligence (AI) is associated with equity-market spillovers in a state-dependent manner and whether AI-intensive sectors are disproportionately exposed during extreme market conditions. We construct a U.S. AI Media Attention Index (US-AIMAI) using a coverage-frequency–based approach that captures daily variation in AI-related news coverage in U.S. media. Using daily log returns from 03 January 2020–16 December 2025 for the KBW NASDAQ Financial Technology Index, the Nasdaq CTA Artificial Intelligence and Robotics Index, the MSCI Semiconductor and Semiconductor Equipment Index, and US-AIMAI, we estimate a quantile-based connectedness framework sensitive to tail dynamics. Results reveal pronounced nonlinearity: connectedness is muted near the center but intensifies in joint tails and across quantile regime shifts. Semiconductor spillovers are stronger and more persistent than those from AI-enabled FinTech. Overall, AI-related media narratives act as a regime-contingent risk amplifier, weakening diversification during heightened market stress.