Tiktok's Polish Elections Labels: Only Sometimes, And Only For Some

Investigating TikTok's inconsistent election information labels and exclusion of Polish diaspora during 2025 Presidential Election coverage.

09-06-2025

Our research examines TikTok's implementation of election information labels during Poland's 2025 Presidential Election. Key findings demonstrate TikTok's failure to meet European Commission guidelines for electoral process protection:

Election labels appeared inconsistently, missing from majority of relevant political content
Geographic discrimination: labels visible only to users physically located in Poland, excluding 20+ million Polish diaspora worldwide
Inadequate response to election fraud allegations: 23 videos with 4.5+ million views spread claims of rigged elections
Nearly 80% of fraud allegation content lacked required election information labels
AI-generated disinformation went undetected: 4 videos used generative AI imagery without mandatory AI disclosure labels

Our exploratory analysis captured content calling to cancel election results and discouraging voter participation and highlights TikTok's inconsistent enforcement of its own policies and inadequate protection of democratic processes, particularly affecting diaspora communities.

Solution & Methodology

Using a non-personalized TikTok account and VPN set to Poland, we systematically searched election-related hashtags on May 26th, 2025. Through snowball sampling, we identified 7 key accounts posting election fraud allegations, analyzing over 1,200 posts since January 2025. We qualitatively coded 23 posts containing fraud allegations from April-June 2025, testing TikTok's election label visibility across Poland and six major Polish diaspora locations (USA, Germany, Brazil, Canada, France, UK). Using 4CAT digital methods toolkit, we analyzed posting patterns and label consistency.

20-11-2024
AI Forensics found 23 videos with 4.3M views spread disinformation about Polish election fraud, with 80% lacking proper election labels and AI-generated content unmarked.