Study: YouTube recommends more problematic content to right-leaning users
A December 2023 study in the Proceedings of the National Academy of Sciences audited YouTube's recommendations using about 100,000 automated accounts. It found the algorithm mostly recommends videos matching a user's ideology. Right-leaning accounts received more recommendations from channels the researchers classed as problematic.

The study complicates claims of anti-conservative suppression in recommendations without showing that any party benefited.
What the record shows
UC Davis researchers led by Magdalena Wojcieszak and Muhammad Haroon ran audits in 2021 and 2022, with sock-puppet accounts watching about 10 million videos from 120,073 channels. They inferred channel leaning from the political Twitter accounts each channel followed. Over 36% of accounts got recommendations from problematic channels: 40% for the most right-leaning accounts versus 32% for centrist and left-leaning ones. Wojcieszak said the algorithm 'does recommend videos that mostly match a user's ideology.' The finding cuts against claims that the recommendation system starves right-leaning viewers of right-leaning content.
In their words
“does recommend videos that mostly match a user's ideology”— Magdalena Wojcieszak, UC Davis communication professor · source
The study measured what automated accounts were shown, not real users, and its 'problematic' and ideology classifications are researcher judgments that critics may dispute.
Sources
- UC Davis: YouTube recommendations lead to more extremist content for right-leaning usershttps://www.ucdavis.edu/curiosity/news/youtube-video-recommendations-lead-more-extremist-content-right-leaning-users-researchers
- PNAS paper (author PDF): Auditing YouTube's recommendation systemhttps://pmlab.cse.usf.edu/assets/pubs/pnas2023.pdf
- Tech Policy Press: New study suggests rightwing bias in YouTube recommendation algorithmhttps://techpolicy.press/new-study-suggests-rightwing-bias-in-youtube-recommendation-algorithm