Work

News, Politics, Societal Issues, and Hate on Youths’ Donated Social Media Feeds

Year: 2026
Type: preprint
Abstract: Youth use social media as a main source for news, political, and societal information, raising urgent questions about the content of such news-like videos, such as the presence of hate speech. Characterizing their actual exposure to news-like content and hate speech requires innovative methods that are not dependent on subjective recall or synthetic accounts. Here, youth (16–25 years old) donated their TikTok feed histories spanning October 2023 till May 2026, resulting in ~6.7M public videos for our analyses. For each video, we extracted the transcripts, descriptions, hashtags, and viewing durations. We then used large-scale automated content classification with large language models (Llama 3.1-70B) to identify videos about news, politics, and major societal issues, such as climate change, migration, and war. We find that nearly all (98%) of the news-like content did not come from accounts that youth followed, suggesting that these were mostly algorithmic recommendations. The prevalence of news-like videos peaked during important political events such as elections. Only 13% originated from established news organizations present on TikTok. Instead, most of these videos came from niche content creators. These niche creators were more likely to produce hate speech videos in news-like content compared to other creators. Hate speech videos most often mentioned specific politicians, political parties, and minority groups. We cross checked the identified hate speech videoIDs with the EU transparency database and found that none were removed by the platform, suggesting inefficient content moderation. Our methods offer a scalable approach to directly measure youths’ exposure to social media content. This provides a foundation for investigating how digital information environments shape societal perceptions and attitudes during adolescence and emerging adulthood.
Language: English
Open Access status: green
OpenAlex ID: W7236466839
Cites:
Cited by:
Related to: 10