A Nature paper published on 6 May found that TikTok’s recommendation system served an uneven mix of 2024 US election videos, with the strongest skew in anti-Democratic content.
The study tracked 323 controlled TikTok accounts over 27 weeks and more than 280,000 recommendations in New York, Texas and Georgia, shifting the focus from what TikTok removes to what its For You feed actually delivers.
The prior argument around election integrity on TikTok has mostly been about moderation: banned political ads, fact-checking partners and misinformation takedowns. TikTok said in a September 2024 Newsroom post that its policies prohibit paid political ads, and that in the preceding four weeks it had removed 43,000 pieces of misinformation content and 2,600 items tied to civic and election integrity.
What the Nature paper examined instead was recommendation output. The researchers classified videos by partisan alignment and found that Republican-seeded accounts received about 11.5% more co-partisan content than comparable Democratic-seeded accounts, while Democratic-seeded accounts were about 7.5% more likely to be shown cross-partisan content. In plain English: one side got more of what it already liked, while the other got more of the other side’s material.
The paper says the asymmetry was not a simple broad conservative tilt. Anti-Democratic videos stood out as the most likely cross-partisan recommendations, with the authors reporting a coefficient of -0.0641 and P < 0.0001. That is the part likely to survive the headline churn: not just partisan imbalance, but imbalance concentrated in content hostile to Democrats.
The issue-specific patterns were uneven too. Democratic-seeded accounts were shown more cross-partisan material on immigration and crime, while Republican-seeded accounts saw more on abortion. The paper also reported that Republican-aligned videos were more concentrated in immigration and foreign policy, while Democratic-aligned material appeared more often in climate change and abortion content.
TikTok disputed the study’s real-world relevance. In a statement reported by The Guardian, the company said: “This artificial experiment with fake accounts does not reflect how people actually use TikTok.” TikTok also argued that users can shape their feeds through follows, likes, searches and other controls.
That objection is load-bearing, because the study used dummy accounts in the early stages of platform use rather than long-lived real users. It also relied on English-language video transcripts, which may miss visual cues and multilingual political signals, and it covered accounts based in only three US states.
The paper does not identify the cause of the skew. As Phys.org noted in its coverage, the imbalance could come from TikTok’s ranking system, from the pool of political videos available on the platform, or from some combination of the two. The survey portion of the study, which included 1,008 US TikTok users, was preregistered and asked about perceived shifts in their feeds, but it does not resolve that causal question either.
TikTok has continued to frame its election work around moderation and context tools. In April 2025, the company said it was testing a US feature called Footnotes and that it works with more than 20 IFCN-accredited fact-checking organizations across 60 languages and 130 markets. The Nature paper points at a different question: whether a platform can enforce election rules and still have its recommendation system lean one way anyway.
The study, “Systematic partisan content skews in TikTok during the 2024 US elections,” is now published in Nature.
Key Takeaways
- A Nature study found TikTok’s recommendations during the 2024 US election skewed toward anti-Democratic content.
- The researchers used 323 controlled accounts over 27 weeks and analyzed more than 280,000 recommended videos.
- Republican-seeded accounts received about 11.5% more co-partisan content than Democratic-seeded accounts.
- Democratic-seeded accounts were about 7.5% more likely to be shown cross-partisan content, much of it anti-Democratic.
- TikTok said the experiment used fake accounts and does not reflect how real people use the app.
Further Reading
- Nature: Systematic partisan content skews in TikTok during the 2024 US elections, Primary research paper reporting partisan skews in TikTok recommendations during the 2024 US election.
- The Guardian: TikTok’s algorithm favored Republican content in 2024 US elections, study finds, Journalism summary with TikTok’s rebuttal and the study’s headline numbers.
- Phys.org: TikTok algorithm showed a pro-Republican bias during the last US presidential election, Explainer coverage noting limitations and the inability to establish cause.
- TikTok Newsroom: TikTok’s Continued Commitment to Combating Disinformation Under the EU Code of Practice, Official TikTok statement on political ads, misinformation enforcement, and election-integrity removals.
- TikTok Newsroom: Testing a new feature to enhance content on TikTok, Official TikTok announcement of Footnotes, a context feature for TikTok content.
TOPIC VOCABULARY (from the research brief, may inform your keyword choice, but the article body is authoritative):
TikTok algorithm, anti-Democratic videos, 2024 US election, Nature study, partisan content