Recommending Hate: How TikTok's Search Algorithms Reproduce Societal Bias
After entering racist/misogynistic slurs, ~two-thirds of examined videos perpetuated harmful stereotypes; across four languages, search and recommendation algorithms connected users to content targeting marginalized groups.
Executive summary
AI-generatedThis summary was generated by AI from the original report to make it easier to scan and cite. It is not a substitute for the source — read the original above.
ISD analyzed TikTok's search functionality across four languages — English, French, German, and Hungarian — to assess how the platform's search and recommendation algorithms handle queries related to marginalized groups, as part of a wider project on online gender-based violence funded by the German Federal Foreign Office around the 2024 European Parliament elections.
After entering racist and misogynistic search terms, researchers found that roughly two-thirds of the videos surfaced across the four languages perpetuated harmful stereotypes, with search and recommendation systems consistently connecting users to content that objectifies or degrades presumed members of marginalized groups. The consistency of this pattern across distinct linguistic and national contexts suggested the bias was embedded in the platform's underlying search and ranking systems rather than isolated to any one market.
The report concludes that TikTok's search and recommender systems reproduce and risk amplifying existing societal biases, and calls for stronger safeguards and mitigation measures from both the company and lawmakers to reduce algorithmically driven harm to marginalized groups.
Think this summary is wrong? Contact us.