A research team at Two Six Technologies developed ML models to identify inauthentic pro-Russia and pro-China accounts on X, tracking behavior from 2024–2026. Key finding: these actors are using AI to improve content quality rather than increase volume. Post counts dropped by half, but image use quadrupled for pro-Russia accounts and doubled for pro-China accounts. Accounts are mimicking human behavior more convincingly — slower posting speeds, sleep-like inactivity periods, and expanded language use (pro-Russia accounts now post in a median of six languages). Despite the quality-over-quantity shift, most inauthentic accounts get minimal engagement, though ~15 pro-Russia outlier accounts with tens of thousands of followers do gain real traction. A major narrative shift was also detected: pro-Russia accounts dramatically increased anti-US and anti-Trump messaging in 2026, reversing prior pro-Trump stances, likely due to frustration over the Ukraine war. Pro-China accounts consistently pushed anti-US narratives and increasingly highlighted China's AI dominance.
