The research, conducted by Germain Gauthier from Bocconi University, involved a randomized experiment with nearly 5,000 active users in the United States over a period of seven weeks, The Conversation reported.
Participants were randomly assigned to one of two feeds. The first group accessed X’s default “For You” feed, which curates posts based on perceived engagement potential, including those from accounts they do not follow. In contrast, the second group utilized a chronological feed that strictly shows posts from followed accounts in the order they were posted.
The results were striking. Users who engaged with the “For You” feed were 4.7 percentage points more likely to prioritize Republican-friendly policies, such as crime, inflation, and immigration concerns. Moreover, these users showed a marked increase in viewing the ongoing criminal investigation into former President Donald Trump as unacceptable.
In another notable trend, those exposed to the algorithm developed a 7.4 percentage point decrease in their positive view of Ukrainian President Volodymyr Zelenskyy, indicating a shift toward pro-Russian sentiments regarding the war in Ukraine.
The algorithm’s effects did not stop at immediate exposure. When users returned to the chronological feed, they continued to follow more right-leaning accounts—a pattern that endured, suggesting the algorithm’s influence has long-lasting repercussions.
This study aligns with previous research showcasing recurring patterns of political bias on social media. Notably, a 2022 study found that X’s algorithm favored right-leaning content over leftist alternatives in six out of seven countries. Further experimental studies in 2025 revealed substantial shifts in user feelings toward their political opponents, a change that the authors noted typically takes years to develop organically within the general population.
