Spammers Are Flooding TikTok with Pro-Conservative Content

This article was originally published by Pivot, an independent, non-profit francophone news outlet based in Québec — PressProgress and Pivot are working together to translate and republish each other’s stories.

Very prolific accounts that publish a large number of comments, sometimes posting the same message thousands of times, are flooding TikTok with pro-Conservative and anti-Liberal content, a tactic that could be an attempt to manipulate the platform’s algorithm. The Canadian ecosystem on TikTok is relatively small, making it more vulnerable to this type of influence campaign.

This is what emerges from an analysis of the TikTok social network conducted by the firm Agoratlas for Pivot.

This analysis examined comments associated with TikTok videos that published content related to the federal election campaign between March 23, the date the election was called, and April 13, 2025.

In total, approximately 1.5 million comments from just over 424,000 users were collected and mapped by Agoratlas.

An account that posts a comment under another account’s video establishes a link between them. These links, in turn, form networks or “communities” of users. Our mapping has allowed us to define four communities: on one side the “Liberals”; on the other the “Conservatives”; between these two communities, the “media” and finally, “Quebec.”

A small number of accounts, mostly members of the “Conservative” community, have published a significant number of comments.

Image: Agoratlas

We detected 928 “spammer” accounts that commented more than 100 times each, for a total of more than 176,000 comments over the course of three weeks. Of these, around 128,000 comments are associated with 600 pro-Conservative accounts.

225 spammers are associated with the Liberal community. They published about 51,000 comments.

In other words, 0.2% of users in our sample are responsible for 15% of all comments, and nearly two-thirds of these accounts (65%) belong to the Conservative community.

Image: Agoratlas

The presence of accounts that publish many comments and post very similar or even identical messages – which is the case here – is a sign of “inauthentic” activity, according to Agoratlas.

Inauthentic activity on social networks is defined as activity that aims to manipulate the platform’s algorithm while hiding the identity of the responsible people or organizations, who remain hidden behind a multitude of seemingly individual accounts.

However, we do not know the true effect this spam campaign has had on the number of views of content published by Conservative accounts.

According to Florent Lefebvre, scientific director at Agoratlas, the Canadian TikTok ecosystem is relatively small compared to those of other countries he has studied, such as France, Germany or Romania.

This small size makes the Canadian ecosystem more vulnerable to influence attempts. Indeed, a small number of inauthentic accounts manage to occupy a significant place, even under some highly followed channels.

This strong presence could encourage TikTok’s algorithm to favour Conservative content and present it to more of its users.

High activity under Conservative content

For some small channels, messages by spammers represent up to a third of the comments received.

The spammers have mainly commented on videos from channels with pro-Conservative messages. While videos associated with the media network have received twice as many views as those associated with the Conservative camp, the latter receive almost twice as many comments.

14 of the 20 channels most commented on by spammers are channels that publish pro-Conservative content, and two are English-language media channels from CBC and CTV.

Image: Agoratlas

It is also noted that spammer accounts have commented on media channel videos such as CTV News and CBC News, which are respectively the second and fifth most spammed channels in total number of comments.

CTV News received around 6,600 comments from spammers (6.7% of total comments on this channel) and CBC News almost 3,700 comments from spammers (7.5% of the total). These two media outlets were mainly targeted by pro-Conservative spammers.

Some accounts associated with the “Liberal” community have also seen a significant percentage of their comments come from spammer accounts. Four of the 20 channels most commented on by spammers are channels with Liberal content by commenters from the Liberal community.

Image: Agoratlas

The presence of a large number of comments under a video could be a signal to the algorithm that there is greater enthusiasm for this content, which can push the platform to present it to larger numbers of users, explains Agoratlas. Spamming comments under a video can therefore lead to an increase in its visibility.

According to Lefebvre, the apparent strategy employed here differs from that observed in Romania and France during recent elections. He noticed the absence of any “raids,” a strategy which was very present during election campaigns in these countries. In December 2024, the Romanian presidential elections were canceled by the country’s Constitutional Court after a declassified intelligence report accused Russia of interfering in the first round of the elections in favour of Călin Georgescu. Georgescu was a little-known far-right nationalist who rocketed to fame through social media manipulation.

The “raid” consists of partisan accounts from one political camp publishing negative or partisan content under videos from the opposing camp or sympathetic comments about their candidate under popular channels. This differs from the strategy of increasing the visibility of one’s own political camp’s content by leaving numerous comments.

Lefebvre gives the example of accounts that, during the French legislative elections of 2024, wrote messages in support of Jordan Bardella, from the National Rally (RN), under videos of popular channels like French YouTuber HugoDécrypte. The objective of this tactic is to make TikTok’s algorithm believe that HugoDécrypte fans are interested in the National Rally to encourage the platform to show these fans more content related to this far-right party.

Lefebvre also notes that in the case of the current Canadian elections, there are no signs of an organized campaign targeting the account of an opposing candidate.

Repeated messages

The studied sample also contains a good number of duplicated comments, meaning copies or very close variants of the same message. Messages found to be duplicated more than 200 times represent a total of 16,000 comments.

A variant of the message “Vote for Carney if you want Canada to become the 51st state of America” was identified 10,000 times. It is by far the most frequently recurring message among the duplicated comments.

Image: Agoratlas

Duplicated content associated with the Liberal community was also found, but in smaller quantities. For example, variants of “Don’t split the vote” published in the Liberal community.

Moreover, 85% of accounts that published duplicated messages are among the “spammers,” those prolific accounts which are mentioned above.

Who are these spammers?

With the data available, it is not possible to know the identity of the people behind the spammer accounts.

The account @al12rogers.com has only one subscriber and has not published any videos. However, this account published up to 120 comments per day during the period studied.

The account @eevir.hanna stands out in our mapping due to the large number of comments its videos received. Yet, it’s an account with a minimal audience by TikTok standards. This channel has 2,668 subscribers and its videos often barely get a few hundred views. (For comparison, the @cbcnews account has 2.3 million subscribers.)

During the period we observed, @eevir.hanna published nearly twenty videos per day, up to 80 on one occasion. Just over 35% of the comments received on these videos, often more than 100 per day, came from spammers.

Pivot also found several accounts on X (formerly Twitter) whose main activity is to relay TikTok videos belonging to accounts from the Conservative community, including those of several “spammers” identified by Agoratlas.

For Agoratlas, the inauthentic activity detected in this analysis appears to be mainly the work of pro-Conservative accounts.

Agoratlas notes that the current Canadian elections “appear much less subject to digital interference than recent elections in Romania or Germany.”

That is to say, “there is indeed manipulation with inauthentic activity,” reports Lefebvre, but it is far from having the same impact as what has been observed elsewhere.

It is therefore less likely to be a state actor or one with similar resources. “It’s comparable with what medium-sized private actors can do to promote a specific narrative or content,” he explains.

Political parties are barely present on TikTok

The official accounts of Canadian political parties are almost absent from the analysis.

This is because neither the Conservative Party of Canada (CPC), nor the Liberal Party of Canada (LPC), nor the Bloc Québécois (BQ) have an official account on TikTok.

The New Democratic Party (NDP) and its leader Jagmeet Singh are on the platform, as is the Green Party of Canada, but their accounts do not stand out in this study among accounts that received a large number of comments.

A representative of the Liberal Party explained to us that they could not comment on their communication strategy during the election campaign, but that the party enjoyed a large audience on networks such as Instagram and Facebook.

The CPC, meanwhile, simply told us, “The Conservative Party of Canada and its leader, Pierre Poilievre, do not use TikTok.”

Distinct TikTok communities

The Canadian sample in our analysis includes 424,477 users who have posted or commented on 1,488 channels that shared content related to the ongoing federal elections.

The research conducted by Agoratlas collected videos from nearly 1,500 channels. These accounts published 42,663 videos during the three weeks studied.

These videos garnered almost 511 million views and just under 1.5 million comments.

The analysis of associated comments highlights four communities, composed of content creators and users who form their audience and comment on their videos.

Image: Agoratlas

The “Conservatives”: This community is strongly mobilized in favour of Pierre Poilievre and displays a virulent rejection of the Liberal Party and Justin Trudeau. It shows great distrust toward the new LPC leader, Mark Carney.

The main accounts that stand out in this sample are @officialmariozelaya, a conservative entrepreneur and influencer based in Ontario and @real.news.media, a pro-Conservative account that publishes video clips related to current events.

The “Liberals”: This community displays a marked rejection of Pierre Poilievre and the Conservative Party of Canada. It shows enthusiasm for Mark Carney and displays discontent with the socio-economic situation.

That said, this group doesn’t only include LPC supporters: accounts from leftist content creators or users close to the NDP are also found here.

“Quebec”: This predominantly French-speaking community displays a rejection of the political status quo. It supports the Bloc Québécois, but also the Conservatives. This category is more eclectic. There is also opposition to the media, political elites and institutions.

Among the channels where the most comments are found are the official accounts of political parties such as the Parti Québécois and Québec Solidaire, as well as media outlets like Qub Radio and Radio-Canada, among others.

The “media”: Comments in this community are highly polarized. There is frustration with Liberal policies as well as discourse favourable to Donald Trump.

The most important accounts in this group are those of CTV News and CBC News: the comments under these media’s content are indeed polarized because supporters from different political camps meet there.

This group also includes accounts like @6ixbuzztv, which started by sharing content related to Toronto’s underground music scene and now publishes polarizing content and has been accused of promoting racist and far-right messages.

Methodology

Agoratlas is a company specializing in social media analysis for marketing and commercial monitoring purposes as well as for socio-political studies like this one.

Agoratlas conducted a search for 113 terms and hashtags in English related to the Canadian election campaign. Some of these terms are generic, such as “CanadaVote,” “Conservative” or “Liberal,” and others are more specific, such as “#Elbowsup” and “#FireTrudeau.” The other terms included in the search are the names of the parties and their leaders.

This search made it possible to compile a list of 7,000 channels that published at least one video containing one of the targeted terms between March 23 and April 13, 2025. Agoratlas then retrieved all the videos from these accounts, totaling 138,000 publications.

Subsequently, all comments published under these videos were captured.

Some of the retrieved accounts and their comments come from accounts based outside Canada, such as NBC News, Fox News or Al Jazeera. This is explained by the use of certain generic terms, such as “Conservative” or “Liberal,” in the search.

For this study, attention was focused on the specifically Canadian community present on TikTok, which allowed for reducing the sample (see the figures presented in the article).

These results do not represent an exhaustive sample of everything published about the Canadian elections on TikTok during the period studied, but the significant amount of data collected makes it possible to detect trends in this ecosystem.

Once the data was obtained, a network study was carried out. Accounts form networks through the comments they publish. By charting these links between accounts, a network map can be created using various algorithms, thus revealing accounts that share a common audience.

The size of the nodes (the points representing user accounts) depends on the number of comments published under their videos. A larger node has therefore received a greater number of comments.

The proximity of nodes to each other depends on the similarity of their audience.

The digital tool D3lta – developed by Viginum, the French service for vigilance and protection against foreign digital interference – was used to detect the presence of duplicated content in comments made by spammers. This tool makes it possible to identify copy-pasted content, translations and rewording.