Selected Papers of #AoIR2023:  

The 24th Annual Conference of the  
Association of Internet Researchers 

Philadelphia, PA, USA / 18-21 Oct 2023 
 
 

 

 

THE IMPACT OF TIKTOK POLICIES ON INFORMATION FLOWS 
DURING TIMES OF WAR: EVIDENCE OF ‘SPLINTERNET’ AND 
‘SHADOW-PROMOTION’ IN RUSSIA 
 
Salvatore Romano 
AI Forensics 
 
Natalie Kerby 
AI Forensics 
 
Miazia Schüler 
AI Forensics 
 
Davide Beraldo 
University of Amsterdam 
 
Ilir Rama 
University of Milano 
 
Introduction 
 
This paper measures the effects of platform polices in times of war, and represents 
different entry-points into understanding how platforms impact information flows (Ciuriak 
2022). We present the outcomes of three tests that assessed content accessibility and 
prioritization on TikTok in Russia in the aftermath of the invasion of Ukraine. The tests 
adopted a combination of scraping and sock-puppet algorithmic audits (Sandvig 2014) 
conducted through Russian-based IP addresses. 
 
This study sheds light on how TikTok adapted to the Russian war censorship laws with 
new policies (TikTok 2022), highlighting the need to closely monitor the platform's policy 
decisions during times of conflict. Using Digital Methods (Rogers 2019) and a cross-
national approach (Gray 2021), we reveal the existence of opaque content removal 
strategies, unevenly implemented policies, and their consequences for Russian users, 
including the spread of pro-war propaganda and their isolation from the global TikTok 
community. Our findings contribute to the understanding of how TikTok policies can 



 

 

contribute to the creation of a 'Splinternet' - a scenario in which countries have vastly 
different information online and separate internet infrastructures (Lemley 2020). 
 
We quantify the changes in the information regarding the war recommended by TikTok 
to Russian users and demonstrate how these changes my have contributed to 
modifying the information diet in the country. Our study also offers insights into the 
methodological challenges of quantifying the effects of Russian techniques that 
international companies navigate when operating in the country (Vendil Pallin 2016). 
 
Test 1: Content Restriction 
 
Our first probe demonstrates how, after the invasion, TikTok restricted access to non-
Russian content in Russia. We started monitoring TikTok's response to the Russian 
invasion with the initial aim of comparing the content t<hat was prioritized to Ukrainian 
versus Russian TikTok users. However, while submitting a query for 'ukraine war' with a 
Russian IP address, we encountered issues with our scrapers. We further investigated 
these issues by conducting tests with a set of automated browsers, querying for a 
number of war-related keywords that were accessible under a Russian residential IP 
address network.  
 
The results of this audit varied over time. At first, the search function on the platform 
was unusable for non-Cyrillic alphabet queries, returning no results. After a few hours, 
the search function was restored for non-Cyrillic alphabet queries. However, we 
observed that the platform had restricted content to Russian-only in both the search 
results and the For You Page (FYP), resulting in a 95% reduction of available content in 
Russia. As a result, accounts created with non-Russian IPs returned an empty page 
when accessed from a Russian IP. We then compared popular accounts accessed 
through a Russian IP address to those created through European IP addresses to 
understand the extent of the content limitations in Russia. 
  
Overall, we found that TikTok was not complying with their own publicly announced 
policies regarding bans on new content in Russia (TikTok 2022). Instead, they made all 
foreign content unavailable in Russia, thereby isolating Russian users from the rest of 
the world. 
 
Test 2: Pro-War Content 
 
Further investigations showed how TikTok unevenly applied their content policies, 
allowing pro-war content to proliferate in Russia despite their claim that they had 
enforced a ban on new content uploads in Russia (Gilbert 2022). To assess whether 
and how new content could be uploaded and made available from Russia, we 
attempted to upload content from a number of accounts and IP addresses—with some 
profiles created in Russia and others not. We repeated these experiments dozens of 
times over twenty days, which allowed us to observe changes and inconsistencies in 
content upload permissions.  
 



 

 

Based on the sample of videos we collected, and after an hashtag snowball process, we 
identified the 12 most prevalent hashtags that were commonly used by both sides to 
express their opinions on the war. 
 
To analyze the content, we automatically accessed around 1,000 of the most viewed 
videos for each selected hashtag and determined which were accessible from a 
Russian IP address. Through our analysis, we were able to identify temporal trends in 
the prominence of pro-war and anti-war content (Figure 1). 
 

 
  Figure 1: Content posted on TikTok from Russian-based accounts  
 
Despite TikTok’s ban announcement on 6 March, we found that the company 
inconsistently applied the ban until 25 March. During the period in which TikTok did not 
comprehensively implement the ban, new content uploads related to the war were 
overwhelmingly pro-war (93.5%), while before the ban was announced, the balance of 
pro-war and anti-war content was roughly equal. Shortly thereafter, new uploads in 
Russia became impossible, and content from outside Russia was banned. Therefore, 
the only information about the Ukraine-Russia war available on TikTok in Russia was 
the content that was uploaded before 25 March, which was dominated by pro-war 
content and propaganda. 
 
Test 3: Shadow-promotion 
 
The third test brought to light a case of “shadow-promotion,” a term we coined to 
describe the algorithmic promotion of content that is supposed to be banned. According 
to TikTok’s own policy updated on 30 June, new content and international content was 
not available to domestic users inside Russia. However, we found that state controlled 
media, political, and non-political content were being promoted to Russian-based users 
via the algorithmically-compiled FYP, even though the videos did not appear on the 
profiles of the users who posted them.  



 

 

 
We created a Russian TikTok profile and tried to upload a video while using a Russian 
IP and non-Russian IP. We then tried to access the video from different locations 
(Russia and Italy) using different access points, for example, accessing the uploader’s 
profile through a direct link, or querying for keywords and/or hashtags. 
 
Subsequently, we explored the availability of content supposedly banned for Russian 
users. We followed accounts with supposedly banned content, and checked the FYP 
afterwards to see if content from the account we had just followed was algorithmically 
recommended. We tested three account types: 1. A Russian media account, e.g.. 
Sputnikvideo, which was labeled by TikTok as ‘Russian state-controlled media’; 2. A 
non-Russian media account, e.g. the official TikTok profile of BBC; 3. A non-political 
account, e.g. the official TikTok account of FCDynamo, the Ukranian professional 
football club based in Kyiv. For all the accounts, we followed the same protocol. We first 
checked their accessibility in Russia, in an EU-country (Italy), and in a non-EU country 
(U.S.) through non-logged browsers. We found that all the supposedly banned content, 
despite not accessible on the users’ profile, did appear in the FYP for Russian users.  
 
Conclusion  
 
Ultimately, these three tests underscore the importance of monitoring TikTok’s policy 
decisions in times of war, which we showcase through geographic-aware algorithmic 
audit methodologies. We were able to demonstrate the existence of unevenly 
implemented policies, and how it risked proliferating propaganda and perpetuating a 
Splinternet.  
 
The Digital Service Act requires big online platforms to publish transparent and detailed 
reports on content moderation at least once a year (art. 13). Although TikTok's 
operations in Russia are outside the scope of EU legislation, as standards emerge for 
platform transparency, its lack of transparency in a high-profile region raises questions.   
 
Our research emphasizes the importance of data transparency regarding TikTok's 
recommendation system and providing civil society access to data to hold the platform 
accountable for its algorithms and content moderation strategies. We have shown how 
this data can help identify biases, discriminatory practices, and manipulative algorithms. 
Further, our investigation stresses the need for social media platforms to have more 
transparent and consistent policies, particularly during times of conflict, to prevent the 
spread of harmful content and the fragmentation of the internet. 
 
References  
 
Ciuriak, Dan. 2022. “The Role of Social Media in Russia’s War on Ukraine.” Available at 
SSRN: https://ssrn.com/abstract=4078863. 
 
Gilbert, David. 2022. “Russian TikTok Influencers Are Being Paid to Spread Kremlin 
Propaganda.” Vice News (03/11/2022). https://www.vice.com/en/article/epxken/russian-
tiktok-influencers-paid-propaganda 
 

https://ssrn.com/abstract=4078863


 

 

Gray, J. E. (2021). The geopolitics of" platforms": The TikTok challenge. Internet policy 
review, 10(2), 1-26. 
 
Lemley, Mark A. 2020. “The Splinternet.” Stanford Law and Economics Olin Working 
Paper #555. http://dx.doi.org/10.2139/ssrn.3664027. 
 
Rogers, Richard. 2019. Doing Digital Methods. 1st edition. Thousand Oaks, CA: SAGE 
Publications. 
 
Sandvig, C., Hamilton, K., Karahalios, K., & Langbort, C. (2014).  Auditing algorithms: 
Research methods for detecting discrimination on  internet platforms. Data and 
discrimination: converting critical concerns into productive inquiry, 22(2014), 4349-4357.  
 
TikTok. 2022. “Bringing More Context to Content on TikTok.” TikTok Newsroom.  
https://newsroom.tiktok.com/en-us/bringing-more-context-to-content-on-
tiktok?utm_source=COMMSTWITTER&utm_medium=social&utm_campaign=030622 
(accessed 23.02.2022). 
 
Vendil Pallin, Carolina. “Internet Control through Ownership: The Case of Russia.” Post-
Soviet Affairs, vol. 33, no. 1, Jan. 2017, pp. 16–33. Taylor and Francis+NEJM, 
https://doi.org/10.1080/1060586X.2015.1121712.