Study Finds TikTok Algorithm Showed Partisan Bias in 2024 US Election Content

A study published in Nature reports that TikTok’s recommendation algorithm disproportionately favored pro-Republican content on the platform during the 2024 US elections in three states: New York, Texas, and Georgia. Researchers from NYU Abu Dhabi created over 300 dummy accounts to simulate typical user behavior by watching videos aligned with either the Democratic or Republican parties. Over 27 weeks, they analyzed more than 280,000 recommended videos. Findings indicated that bots exposed to pro-Republican content saw about 11.5% more content aligned with their views, while pro-Democrat bots were shown 7.5% more pro-Republican content. Pro-Democrat accounts received more anti-Democratic material, whereas pro-Republican accounts were exposed to less anti-Republican content, revealing a partisan skew. The types of issues highlighted varied: Democratic bots encountered more cross-partisan content on immigration and crime, while Republican bots saw more on abortion. TikTok responded that the study, based on artificial accounts, does not reflect real user behavior, highlighting that users control their feed through various tools. Despite these limitations, the study raises concerns about algorithmic influence on political content exposure and its potential impact on young voters, especially given that TikTok’s For You page is largely algorithm-driven and less influenced by user choice.