The next speaker in this session at the Weizenbaum Conference is Felix Gaisbauer, whose interest is in pathways towards engagement with political content on TikTok. The platform has increasingly been identified as an important space for such engagement, with right-wing and far-right actors apparently especially active. This has caused some commentators to call for more non-extremist political content on TikTok, which assumes that such content does not already exist, that there is demand for it, and/or that the TikTok algorithm privileges extremist content.
To better understand this, we do need to distinguish more properly between the supply of and demand for political content on TikTok, though. First, who produces political content for TikTok; second, what content is algorithmically supplied to audiences; and third, what demand for which political content is there actually amongst TikTok users? How are these processes interrelated?
There have been some early studies of political experiences on TikTok already; one German study using bots showed that far-right AfD content was recommended more often to those bots than any other content, though another found that this was true only for right-leaning users. The present study therefore draws on data donations from two initiatives, including a collaboration with news outlets in the context of the 2025 German federal election.
Such data donations provide a great deal more detail on user interests, activities, and interactions with videos on the platform, and thereby capture the demand side much more effectively; they also offer some further information on the algorithmic supply side. To understand the supply of political videos by producers, it is still necessary to scrape the accounts of identified political accounts, however. Data donations are also necessarily skewed by which users are willing to donate their data.
AfD-affiliated accounts turned out to be considerably more active in producing political content; these are not necessarily official party accounts, but also ordinary supporters. Other parties and their followers became more active only around the 2025 election.
Algorithmic supply largely mirrors users’ own political leanings; users predominantly see content from ‘their’ parties, and from others on the same side of politics. Whether such recommended videos are watched fully by users does not seem to be affected by their party affiliations, however; even AfD voters sometimes watched Greens videos in full.
Overall, though, most of the content users encountered tended to be non-political. While political content creation is highly skewed, this is not reflected in algorithmic supply or user engagement patterns.
