Weekly Update 6 | Friday, May 2
The weekly update on the Canadian election provides information on trends and dynamics observed in the information ecosystem (Snapshot), identifies and shares insights on information-related incidents and emerging threats (Incidents), and shaeducational content, research findings and other relevant outputs from our Coalition on Information Ecosystem Resilience (Update). All facts and figures are taken from an original survey and social media data collection by the Media Ecosystem Observatory and the analysis reflects the period from March 23 April 28.
Key takeaways
We open this final weekly update with six reflections from the research team. These are preliminary findings and impressions. Our post-election analysis will have richer survey and social media data that will allow us to more fully assess these reflections as well as the overall state of the information ecosystem during and in the immediate aftermath of the election.
Misinformation powered by generative AI: Our monitoring identified significant misinformation across social media platforms, with numerous instances of generative AI and automated bot activity. AI-generated fake news and deepfakes about political party leaders mimicking outlets like CBC and CTV spread widely on social media during the election. This challenge is exacerbated by Meta's ongoing restrictions on news content in Canada.
Political and platform segregation increasing: The conversation and user-base of social media platforms is becoming increasingly politically segregated. Liberal-affiliated users and candidates are migrating to platforms like Bluesky, while conservative voices are increasingly dominant on platforms like X. Politician usage of TikTok continues to be marginal, although Canadian news outlets have been successful on that platform. This digital segmentation risks exacerbating polarization, potentially deepening political divides and complicating efforts to counter misinformation.
U.S. influence played an important role: While Trump, tariffs and threats of annexation were key issues throughout the campaign, we were specifically concerned about covert U.S. interference in the election. Many Canadians shared this concern. However, despite high levels of public anxiety, we observed low levels of direct involvement or impact from U.S. state actors or important influencers during this election period.
Limited foreign interference: Allegations of foreign influence were a major concern throughout the campaign period. We did not, however, uncover evidence of impactful foreign interference in this election cycle to-date, although we note some methodological limitations. There were minor yet discernible attempts by foreign entities to produce and promote existing misinformation trends and amplify divisive content, particularly targeting diasporic communities. These attempts generated limited engagement and attention in Canada, but their existence underscores the need for continued vigilance regarding potential foreign interference.
Suspicious accounts targeting Mark Carney: On the social media platform, X we observed numerous suspicious accounts targeting political leaders and especially Mark Carney. We continue to be concerned about information manipulation on the platform and observe that a decreased emphasis on election integrity and the general loosening of moderation restrictions are likely contributing to an increasingly adversarial and polarized information environment.
Widespread election integrity misinformation: In the second half of the campaign, we observed concerted efforts to promote claims suggesting the electoral process is inherently compromised. While the immediate impact of such claims appears limited, with trust in the election process remaining high among supporters of all major parties, notable misinformation incidents have gained significant traction online, particularly on platforms like X and TikTok with numerous poorly sourced claims of voting fraud. The increasing visibility of these narratives in Canadian elections is worrying and underscores the importance of continued efforts to promote transparency and trust in our democratic processes.
SNAPSHOT
Election rewind: a look back at the information incidents that marked the campaign
During the election campaign period, the Observatory monitored the media ecosystem to identify and assess potential information incidents that could disrupt Canadian public opinion, voting behaviour, or the electoral process. We define information incidents as stories or events that impact the regular flow or integrity of information, leading to potential or actual harm to the public, government, democracy, or the broader information ecosystem.
Overview of the election period
Prior to the writ drop, we conducted an in-depth, comprehensive* environmental scan of the Canadian information ecosystem to identify key vulnerabilities and potential threatening narratives that could affect the process of the election. We anticipated incidents to emerge related to five key issue areas:
Foreign influence and interference (U.S., China, Russia, and India; concern about U.S. interference surged following the increase in 51st state comments by President Donald Trump);
Diaspora communities (transnational repression, coordinated disinformation campaigns especially for Chinese and Indian communities);
Election integrity (concerns about election fraud, safety of elections technologies, controversy over election results);
Immigration (often linked to themes of affordability, housing, and crime);
Climate change (carbon tax and natural resource extraction);
During the election, some of these issue areas manifested more than others. At the start of the election, the online conversation around foreign influence focused largely on the United States and China, with some limited attention to India. While concerns about these countries remained steady during the 2025 campaign (see Figure 1), concern about the U.S. surged past China and Russia during the election, reaching nearly 70% (+30% increase in one month).
* Methods included topic modelling of in-house digital trace data, literature review, analysis of research, policy, and government reports, as well as lessons learned from previous elections monitoring activities.
Figure 1: Level of concern about countries trying to influence Canadian politics and elections
Amidst foreign influence narratives, some discussion focused on the targeting of diaspora communities, e.g. Chinese influence through RedNote and WeChat. Once early voting started, however, public concern shifted to topics related to election integrity, i.e. concerns about fraud, questions about polling, vote tabulator machines, the ability for election workers to tamper with ballots, among other topics. Both immigration and climate change ended up being topics given little focus this election, and no incidents were associated with either.
Summary of incidents
We identified and investigated 36 information incidents over the course of the election. Less than half of the potential incidents we detected were escalated – 13 minor (Level 1), and 2 to moderate (Level 2) incidents (see here for incident thresholds). Figure 2 shows the cumulative count of potential incidents by week of the election.
Figure 2: Total incidents assessed over the period of the elections
The types of incidents identified spanned 4 categories:
Foreign influence and interference: efforts by external actors to influence domestic discourse;
Election integrity: concerns about the electoral process;
Platform manipulation: use of bots or malicious techniques to manipulate platform engagement; and
Content manipulation (Gen AI): use of altered content to misrepresent reality.
Overall, 1 in 3 incidents identified related to potential foreign interference. Platform and content manipulation (bots and Gen-AI) made up 14% of incidents. As explained later in this update, we identified many threatening claims about election integrity but group these into one overarching incident.
Among various incident types assessed during the 2025 federal election, dis/mis/malinformation incidents steadily increased, peaking in the final week, while foreign influence incidents were more concentrated earlier in the period.
The remainder of this snapshot provides a high-level summary of the incidents identified and reported on during the election period. We will provide a richer summary of incidents in a final post-election report.
-
Four countries - the US, China, India, and Russia - were widely seen as the main foreign actors threatening the election. However, only the United States and China were associated with incidents during the election.
United States: Concerns about the United States surged after discussion of the 51st state started increasing. We reported on two incidents related to this topic (a suspicious pro-51st state Facebook group with high engagement and an online movement supporting annexation of Alberta) for connections between these groups and American sources. We determined that these initiatives were not backed by American organizations nor had significant interference or disruption in the larger Canadian media ecosystem.
China: There was significant online discussion of Chinese ties to the Liberal Party during the election. We flagged numerous claims throughout the election period. Most notably, we investigated an incident reported by SITE’s task force involving a China-based WeChat channel attempting to influence Chinese Canadian communities. After monitoring the channel flagged by SITE, we found very limited Canada-related content, with the most recent post reporting on the election outcome. No misleading narratives or unusual activity aimed at influencing Canadian public opinion were observed. We found that discussion about Joe Tay is tightly restricted on platforms such as WeChat and RedNote, limiting the candidate’s visibility and public discussion about him amongst transnational diaspora communities. We also noted the circulation of partisan narratives on RedNote were predominantly supportive of the Conservative Party and critical of the Liberal Party.
Although we didn’t report on any minor incidents from India and Russia, their actions were reported elsewhere during the election. Reports on Indian interference in the 2022 Conservative leadership race, for example, sparked a wave of online debate that, regardless of coordination, fueled doubts about election integrity and laid the groundwork for future fraud claims. Meanwhile, Russian media focused on themes like U.S. threats, immigration, globalism, Gaza, and elitism, with coverage largely spread via social media. While they skipped the discussion of Mark Carney's ties to China or his offshore assets, they posted on Carney's stance against the Freedom Convoy and alleged Liberal attacks on the People’s Party. DisinfoWatch reported that Russian state media and affiliated accounts targeted Mark Carney with negative narratives portraying him as a global elite and Ukraine supporter, in contrast to Chinese efforts that appeared supportive.
-
Questions around the integrity of the election emerged in high-volume in the lead up to election day. With narratives ranging from claims that ballots marked in pencil may be tampered with or that polling data is manipulated, the surge of claims, level of engagement and perceived threat to the electoral process triggered declaration of election fraud as a moderate level incident. We have investigated and provided more detailed findings on this incident at the end of this update (in the incident section).
More broadly, it’s worth noting that some electoral fraud narratives echoed claims made regarding the B.C. provincial election in fall 2024, including allegations that Elections Canada (like Elections BC) was biased toward the government in power (Liberals federally or NDP in B.C.). Analysis by Chris Tenove, Nishtha Gupta, Mahnan Omar, and Grace Lim in Maclean’s compared three categories of narratives challenging election fairness: 1) misinformation about the voting process (which should be debunked), foreign interference (which warrants continued vigilance), and 3) broader criticisms of our polluted and polarized information system (which deserves public discussion and action).
-
There was a notable volume of suspicious activity on X and TikTok throughout the election.
On X, the DFRLab observed many instances of bot-like accounts that excessively posted repeated images attacking politicians, mostly targeting the Liberal Party and Mark Carney. We investigated a suspect bot campaign pushing a narrative of Carney being connected to Jeffrey Epstein. We found potential bot accounts amplified the story but had little impact, and we suspect the spread was most likely due to commentators on X sharing the image in early 2025. We are also aware of reports by ResetTech, the Financial Times, and CAAD on potential bot networks on X spreading pro-Russian and right-wing narratives, some of which were critical of Mark Carney and supportive of Pierre Poilievre. In our preliminary assessment, we determined some suspicious and seemingly coordinated activity but found generally low level of engagement with the suspicious accounts. These reports provide evidence that there were attempts to use automated accounts to manipulate user opinion on X.
On TikTok, according to Pol Comm Tech Lab, nearly 30% of over 1000 accounts identified during the campaign had no face or AI generated images in their profile pictures and posts. These accounts had small followings, but their content reached election-focused For You Pages. It is not clear if these accounts were authentic or alternatively coordinated, paid or otherwise compensated. While we didn’t declare a specific incident on TikTok and bots are hard to detect, we did see significant bot activity from the large numbers of accounts posting thousands of posts and comments per hour. Looking at election related TikTok content, however, the most active posters, with over one thousand election related posts, did not show indicators of bot activity.
-
Although altered and deepfake content were largely used for humor, fraudulent political ads featuring doctored images of party leaders and fake news broadcast content contributed to the perception of a polluted information environment.
Deepfake news and ads: We flagged a moderate (Level 2) incident for AI-generated ads masquerading as legitimate Canadian news sources and linked to cryptocurrency scams. The high volume of ads (nearly 100 pages and reaching hundreds of thousands of Canadians or more) posed a severe risk of influencing public perception during the election period. The prevalence of these ads might also suggest that success online looks more like running a crypto-scam than creating a good news outlet, raising doubts about platforms’ effect on media economics that become crucial during elections.
Deepfakes of politicians: While convincing deep-fake content was rare, we observed a number of realistic and more consequential deepfakes including AI-generated images of Mark Carney with Jeffrey Epstein and Ghislaine Maxwell. The use of deepfake content of Party leaders, particularly in fake ads masquerading as legitimate news, can dilute the quality of the Canadian news cycle and journalism crucial during electoral campaign periods.
Fake accounts and AI-generated content: In collaboration with The Pol Comm Tech Lab, we encountered some accounts that appeared to use fake avatars on X and TikTok that would push anti-Liberal and pro-Trump content and AI-generated content. The use of fake personas and synthetic media to promote partisan narratives, particularly when done covertly, can erode trust in the information environment, distort authentic public discourse, and potentially constitute a form of information manipulation or interference.
The Applied AI Institute highlights elections can suffer because of bigger changes in the media ecosystem. The deluge of AI-assisted advertising shows how poor ad moderation can undermine the quality of information during elections. More so, the prevalence of these ads might suggest that success online looks more like running a crypto-scam than creating a good news outlet, raising doubts about platforms' effect on media economics that become crucial during elections.
INFORMATION INCIDENTS
During the election, we reported on information incidents that could mislead the public and disrupt the democratic processes. We are primarily concerned with covert information manipulation and foreign interference efforts as opposed to instances of influence which are definitionally overt and public. This week, we report back on our investigations into two moderate incidents. Click here to learn more about our incident response thresholds.
Moderate ongoing incidents
AI-generated fake news
Since early in the election, we have been investigating fraudulent political ads masquerading as legitimate news sources on social media (primarily on Facebook, but also on other major social media platforms). The proliferation of these ads was classified as a moderate incident given the scale, the political nature of the content, the misinformation spread, and the potential to influence the attitudes of Canadians. Based on observations from our boots-on-the-ground team, tips from the public, and keyword searches on social media, we identified nearly 100 Facebook pages that purchased these fraudulent ads, but the actual number of fake ads is likely to be substantially larger. A large number of Canadians encountered this content: 24% of Canadians believe they were exposed to social media posts and websites posing as legitimate news sources by imitating their name, logo, or design during the campaign.
We estimate that the ads with the highest levels of engagement (around 6,000 reactions, comments, and reshares) were viewed by more than 100,000 Canadians. The ads and fake articles were highly political, featuring deepfakes of party leaders and inflammatory political rhetoric. The content was not labeled as political on Facebook and circulated on Google/YouTube, even though political advertising is not allowed on those platforms in Canada. These ads appeared to be less prevalent after Election Day: none of our team members encountered them and no screenshots were sent to us via the Tipline. However, more than 1/3rd of the Pages we identified have not been removed by Meta, and some have started targeting citizens of other countries (primarily in Europe) with similar scams. We provide more details on the ads and fake articles, including their origin, potential impact, and Meta’s response, in these two incident updates we published during the campaign.
Surge in claims actively undermining election integrity
Last week we flagged a surge in claims actively undermining election integrity as a moderate incident given a sizable increase in their prevalence and reach. This week we report back with additional details on claims around election integrity and their evolution over the course of the campaign and in the immediate aftermath of the election. All the analysis presented here is preliminary, with additional investigation to come in the post-election period.
During the election period (March 23 - April 28), we identified common election fraud claims observed by our boots-on-the-ground team and research partners, analysed mentions of those election fraud topics in our dataset of prominent Canadian voices online, and further assessed the volume of claims on X where the most content on these topics can be found.
Which claims were the most prominent online?
The most prominent narratives of election fraud flagged during the election cycle were:
Ineligible or fraudulent votes being cast: claims that non-electors were able to vote; individuals going to vote but finding their name had already been crossed out or that they were not registered; being able to vote with any forms of ID (e.g. prescription bottle); people voting at multiple polling stations
Ballot manipulation: Elections Canada throwing away ballots with misspelled names; counting ballots early; workers taking unsealed ballot boxes home; insecure handling of ballots at polling stations; discoveries of ballots in dumpsters.
This includes a sub-narrative of “bring your pen” urging electors to bring their own pens instead of using the provided pencil so their vote can’t be erased. Out of all election fraud narratives, this one received the most social media traction.
Fake polls: questioning the validity of polls, use of generative AI content to discredit polling firms.
Carleton’s long ballot: speculation for why Elections Canada permitted 91 candidates on the ballot for Carleton, Pierre Poilievre’s riding.
When did these claims gain prominence online?
In our dataset of social media posts of prominent Canadian voices online, we identified posts where each of these claims were mentioned during the election cycle (in this preliminary analysis, we do not differentiate between election fraud being claimed and counterclaims against it). Our analysis showed that the majority (67%) of the posts around election fraud are posted on X. However, we find that the majority (71%) of engagement, which is measured by likes, happens on video based platforms (i.e. YouTube and TikTok) despite far fewer posts. We find that election frauds topics followed different timelines:
Topics related to the voting process, such as “bring your pen” and ineligible votes, peaked in engagement around the advanced polling period (April 18-21) and showed signs of increase right before Election Day.
Some topics emerged independently from key election dates, such as claims around fake polls, whose increase in engagement coincided with the increase in media coverage on Conservative rally banners sporting the words “Do you believe the polls?” around April 10th.
Discussion around ballot manipulation increased in the week leading up to Election Day.
Figure 3: Prevalence of main claims questioning election integrity throughout the 2025 Canadian federal election
What has happened to the online conversation post-election?
After Election Day, our boots-on-the-ground team used “avatar” accounts (new profiles specifically calibrated to appear as left-leaning, centrist, right-leaning, and non-political users) to review the most prevalent narratives circulating on TikTok, Facebook, Instagram, X, Bluesky, and RedNote. All the main allegations of election fraud identified during the election period resurfaced as popular claims. In particular, there was a focus on Pierre Poilievre’s loss in Carleton, with the long ballot topic evolving into new election fraud claims alleging issues with early counting specially allowed for this riding and unfair redistricting. These narratives were almost entirely driven by right-leaning accounts, but nevertheless appeared on feeds of accounts calibrated to appear as centrist or apolitical.
This was corroborated by our preliminary post-election analysis on X/Twitter, where we isolated election fraud claims from general discussion of the topic. We saw an explosion of posts pushing election fraud narratives on April 29th, reaching a total of 813 posts and constituting an 1667% increase from a daily average of 46 posts pre-election. Among the 459 election fraud claims that assigned blame, the majority targeted Carney and the Liberals, but claims also focused on foreign countries including China and the United States. We anticipate these claims to continue in the near-term. We do not think that these claims will be amplified across the media ecosystem in the long-term, but it may still impact public trust in electoral and democratic institutions and the incoming administration.
We will continue our investigation in a post-election report.
Did these claims influence trust?
While there was considerable discourse online, we continue to see stable levels of trust amongst Canadians in the election results and in Elections Canada’s ability to fairly administer the election. As with the previous week, our surveys reported consistent levels of trust (74% reporting “A fair amount” or “A lot of trust”) in the election results and confidence in Elections Canada to run the election fairly (85% reporting “Moderate confidence” or more).
COALITION RESOURCES
Digital Forensic Research Lab (DFRLab)
DFRLab’s latest piece, “How Social Media Shaped the 2025 Canadian Election,” analyzes how today’s information ecosystem—shaped by weak platform enforcement, AI-generated misinformation, and foreign interference—reveals evolving challenges to electoral integrity.
Information Integrity Lab
On May 7, the Information Integrity Lab is hosting “The Arctic and Climate Change: The Intersection of Geopolitics and Disinformation”. The conference will explore perspectives from across the Arctic states on state-based malign influence campaigns aimed at distorting or undermining climate debates for strategic gain. To register, click here.
THIS WEEK FROM THE TIPLINE:
We have received a total of 230 submissions during the election period including 17 new submissions this week.
59% of the new submissions involve posts on Facebook, 24% from YouTube, 12% from News Outlet/Websites, and the remainder of 6% Instagram.
The submissions form this week focused on 2 dominant topics*:
Users report content spreading misleading claims about elections, often involving Trudeau, CBC, and manipulated narratives.
Users highlight the spread of false claims about Carney across Facebook, emphasizing fabricated links and misleading narratives.
*Note: The topics submitted reflect common user concerns; while some may lead to potential incidents within the media ecosystem, not all necessarily indicate imminent threats or events.
We would like to thank everyone who submitted to the Tipline. Whether big or small, your contributions are deeply appreciated. Even though the elections are over, the Tipline will remain open and we will continue to review new submissions as we monitor evolving threats within the information ecosystem.