Thank you for the opportunity to testify before this committee today regarding the threat of foreign interference in the 2025 federal election. I'm especially pleased to offer my perspective as director of a U.S.-based research centre that counters online foreign interference and someone who has spent years working to uphold Canada's digital sovereignty, most recently as an associate member of the Canadian Digital Media Research Network.
I've overseen several election defence efforts this past year, including in France, Moldova, Poland and the United States. Canada's was unique. Canada is not the nation most targeted by foreign interference, but it has become a clear test bed for AI-enabled manipulation, and it is especially vulnerable to changes in the policies of large social media companies. Rules set in Silicon Valley have a clear and disproportionate effect on the Canadian public.
Let me give you an example. In July, the DFRLab identified 42 AI-generated YouTube channels posing as Canadian news sources and focusing extensively on allegations of voter fraud and especially on Alberta separatism. These channels used low-quality AI imagery, AI voices and AI scripts. One video, titled “Mark Carney Tears Up As Governor General Makes Shocking Ruling On Election Results”, received 220,000 views—a shocking number given the specificity of the topic.
I cannot say that this was a Russian or Chinese network. In fact, given the adjacency of these assets to Canada-related cryptocurrency scams, it may well have been financially motivated. What I am certain of is that this network—and there are many other examples like it—had more impact on Canadian public perception than activity by the Youli-Youmian WeChat account, which was the largest known state influence effort this election cycle, as attributed to them in April by the SITE task force.
Moreover, YouTube was slow to take action against this election delegitimization network and only did so after revising its policies on AI-generated content. It did not take action against this network because of any violation of its election integrity policies, although such a violation had almost certainly occurred.
Other platform policies compound these issues. If you were browsing Facebook in the lead-up to the election, the so-called Meta news blackout would have kept you from finding useful information from the CBC about election safeguards. Instead, you would have found AI-generated slop from the network I just discussed or even more damaging falsehoods. Our study of the spread of another YouTube video on Facebook, which falsely claimed to show Canadian poll workers stealing ballots, created a narrative that bled into X and generated at least 32,000 mentions and 135 million impressions on X between April 27 and April 30.
These were not exactly foreign influence operations. They were certainly the narratives that foreign actors found most compelling and would therefore amplify. Among foreign observers of the election, they also resonated disproportionately with Canadian election watchers in the United States. As this audience engaged with them, the narratives were amplified by algorithms of predominantly U.S.-based social media companies and served to a Canadian public that, in some cases, had been banned from accessing their own news sources.
Finally, because I've spent years studying U.S. election protection systems and Canada's parallel development of the rapid response mechanism and SITE task force, I want to conclude by acknowledging the judicious approach of Canada's election defenders. Even when you've identified clear state-linked activity, it is often the case that acknowledgement of that activity can have a worse impact on public confidence than the activity itself.
I believe that Canada's approach to public disclosures in 2025 compared favourably to that of the United States in 2024. Emerging election defence efforts in the European Union under the democracy shield would do well to look to Canada's example.
Thank you again for the opportunity. I look forward to your questions.