It starts with a feeling many of us recognize: you scroll through social media, see a political opinion you find outrageous, and wonder whether a real person actually believes this—or whether some shadowy operation cooked it up in a server farm halfway around the world. In Canada, that suspicion has become almost reflexive. But Aengus Bridgman, a digital misinformation researcher and director of McGill University’s Media Ecosystem Observatory, is warning that Canadians need to resist turning that reflex into a weapon against their real-life political opponents. The temptation was never stronger than after Ukrainian missile strikes reportedly knocked out Russian data centres in October, triggering a wave of claims that online bot activity—and the toxic political content it amplifies—had suddenly collapsed. It was a seductive story: right-wing activity plunging, hateful comments drying up, social media becoming “a little easier to breathe.” Bridgman understands why people want to believe it, but he cautions that this kind of thinking is dangerous. “Sometimes these moments can actually be very helpful because they can give us a moment of self-reflection,” he told The Hill Times. “As a country, we have become far more aware of the threat of foreign interference, but it’s now leading to these moments of false concern, and we need to be careful not to cry ‘wolf.’” The problem, he argues, is not acknowledging that foreign interference exists—it is using that concern as a cudgel to dismiss legitimate political divisions, to label fellow citizens as un-Canadian, and to avoid engaging with ideas we dislike.
The saga began with real events. From Oct. 8 to 11, Ukraine launched a series of missile strikes against computing infrastructure at Yandex facilities in Russia. News reports documented significant damage, and Cloudflare, a web security company, recorded a 94 per cent decline in automated server traffic from those data centres. That technical observation quickly got mangled in the fever swamp of social media. A chart shared by Tomáš Zdechovský, a Czech member of the European Parliament, suggested that activity around the far-right Alternative for Germany party had plunged after the strikes. “If confirmed, this is telling: Russian data centres get hit, and online activity around AfD suddenly plummets,” he wrote. The post went viral, accumulating millions of views. Ukrainian President Volodymyr Zelenskyy amplified the sentiment at the Warsaw Security Forum, saying that social media had seen “less hate, less division, and fewer provocations” since the strikes, and that there were “fewer bots.” His office later clarified that he was referring to the general atmosphere on social media rather than hard data about bot activity, and pointed to reports of fewer provocative and hateful comments. But the narrative was already loose in the wild. Similar claims popped up about lower engagement for accounts linked to the UK’s Reform Party, pro-MAGA accounts in the United States, and in Canada, accounts promoting Alberta separatism or supporting Conservative MPs. The implication was clear: if your political opponents lose engagement, they must have been bots all along. Bridgman, however, reframed what was actually being measured. At face value, baseline bot activity from that particular data centre did go down—because the infrastructure was physically destroyed and could no longer perform the automated functions required to crawl the web. But that is not the story that captured imaginations. “The story that got picked up and circulated kind of en masse was that the internet is suddenly a better place,” he said. That story was compelling enough to ignore the nuance.
Bridgman and his team at the Media Ecosystem Observatory decided to test the claim. Their Oct. 15 analysis examined social media posts from 113 prominent X accounts between Sept. 14 and Oct. 11, focusing on the Alberta referendum debate. They found no significant dip in attention toward either pro- or anti-separatist accounts after the Ukrainian strikes. What they did find was a meta-pattern: posts endorsing the claim that bot activity had collapsed generated far more attention than posts disputing it, even though both sides produced a comparable volume of content. This, Bridgman said, is one of the best recent examples that truthfulness is only the tiniest fraction of what drives engagement. A previous MEO analysis from September had also found no evidence of effective Russian or U.S. influence on social media targeting Alberta’s referendum. None of this means foreign interference is a myth—Russia is undoubtedly running campaigns somewhere, and the absence of proof is not proof of absence. But the specific narrative that took hold was not supported by the data. Bridgman said he understands why the story was so appealing. The idea that “the digital sphere would be a healthy, lovely, non-polarized place without Russia” is intoxicating. It offers a clean explanation for polarization, division, and the proliferation of ugly opinions. But it also, he warned, “offloads all responsibility to this evil other and allows you to ignore what is actually going on.” Canadians have seen this before, with the Freedom Convoy and the anti-vaccine movement, and now with Alberta separatism. “We’re seeing people want to believe their political opponents are not real,” he said. That desire may be human, but it is also a form of misinformation. Bridgman found no indication that the viral claims were being spread with malicious intent. They were not necessarily part of a foreign plot; they were part of a natural, deeply human tendency to embrace flattering stories. But good intentions do not make something true.
This is not just a left-wing or right-wing problem, though it often wears a partisan costume. Bridgman pointed out that accusations of astroturfed bot campaigns and foreign interference have become a convenient way for both sides of the political spectrum to dismiss their opponents as “No True Canadians.” If someone supports an idea you despise, it is easier to claim that their support is manufactured from abroad than to confront the possibility that real people, with real grievances, genuinely disagree with you. During the last federal campaign, Bridgman recalled, there were accusations that support for the Liberals online was boosted by China, and that Conservatives were being propped up by Russia. More recently, during Quebec’s provincial election, a Coalition Avenir Québec leader suggested that Americans might support the Parti Québécois to destabilize Canada—before admitting there was no evidence for it. It backfired, but it was another example of the same impulse. The difficulty is that foreign interference is not imaginary. Canada has spent years, including through a public inquiry, examining the reality of foreign meddling. Bridgman wants that threat to be taken seriously. “You do want genuine foreign interference to be kind of detected and discarded,” he said. But the question is how to make the population resilient to manipulation without over-indexing on fear. If every online disagreement is treated as evidence of a foreign plot, then the actual conversation Canadians need to have about their own society becomes impossible. Bridgman’s proposed antidote is not to become more suspicious, but to build a healthier, more robust, and more de-anonymized online public square. That means accepting that not every awful opinion is a bot, that not every viral trend is a conspiracy, and that the person on the other side of the screen might actually be your neighbour. It also means being honest about your own side’s susceptibility to misinformation, rather than assuming that only the other side gets fooled.
The Alberta referendum provided a particularly clear case study. The Media Ecosystem Observatory’s September analysis tracked more than 100 prominent voices on social media discussing the question of Alberta independence. Of those, 66 argued for an independent Alberta, while 29 pushed for Alberta to remain in Canada. The pro-independence side dominated engagement by a staggering eight-to-one margin. Canadian accounts posting about Alberta independence received 1.7 times the engagement of accounts discussing any other political topic, and the largest independence influencers received 3.8 times that engagement. AI-generated content also outperformed human-created images by more than three to one. On the surface, this looked like a massive wave of real or manufactured enthusiasm for separation. But Bridgman cautioned against reading too much into it. He said neither the September analysis nor the October follow-up proves the absence of foreign influence campaigns targeting the referendum; Russia would likely run such campaigns using easily transferable servers, and any traffic from them would be designed to look like ordinary users. But for the most part, the loudest and most prominent voices were identifiable people. “Even if you disagree, they’re real,” Bridgman said. Much of the observable difference in engagement between the two sides likely reflected platform algorithms, as well as differences in where each side chose to focus its energy. The independence movement, it seems, was simply more passionate about showing up online. Yet that passion did not translate into votes. Preliminary results from the Oct. 19 referendum showed Option 1—remaining in Canada—received more than 1.5 million votes, while Option 2, calling for a second binding referendum on separation, received just over 519,000 votes. The side that dominated online engagement lost by roughly three to one. Nathaniel Arfin, an AI and IT security expert, said the lesson should be clear: the prominence of a social media narrative is not evidence of authenticity or popular support. “That’s something I think these groups have to reckon with: being the loudest isn’t the same as being able to rally support,” he said. An “endless cannon” of content, no matter how well-financed or algorithmically boosted, does not necessarily move real people to action. The internet is not the country, and a trending hashtag is not a mandate.
There is, of course, a real risk that the underlying threat does not disappear. Whatever the Ukrainian strikes disrupted, bot networks using openly available AI models can be relocated. Arfin argued there is still a genuine concern over Russia’s capacity to establish data centre “bullshit factories”—facilities dedicated to manufacturing fake content at massive scale. The ability to lie on the internet is nothing new; the volume and sophistication of fabricated content have changed. “People can say anything on the internet as loudly as they want, but turning up the volume doesn’t equal a growing movement,” Arfin said. That is a crucial reminder in an age of algorithmic amplification. But Bridgman also found something almost poetic in the irony of the whole episode. The people most likely to spread the story that Russian bot activity had vanished were often on the left, the very cohort that prides itself on fact-checking and fighting disinformation. They were tricked by a narrative that flattered their worldview. “Liberals and the ‘remain’ side are always very quick to point out misinformation, but, in this moment, they were tricked by it to a certain extent,” Bridgman said. He hopes the episode can build empathy rather than smugness—an understanding of how difficult these online dynamics truly are, and how platform propagation of content is genuinely agnostic to the truth. We all want to believe that our political opponents are not real, that their opinions are manufactured, that the country would be united if not for foreign saboteurs. But Canada’s political divisions are real, and they will not be healed by dismissing them as bots. The best defence against foreign interference is not paranoia or the weaponization of concern, but a healthy public conversation where people are allowed to be wrong, allowed to be different, and still recognized as fellow citizens. That means accepting that some of the voices we find most objectionable are actually human beings. It means doing the hard work of argument rather than the easy work of accusation. And it means remembering that whenever the internet suddenly seems like a better place, the explanation might be more complicated than we want it to be. The truth is rarely as clean as a viral chart, and the fight for a healthy democracy cannot be outsourced to a missile strike or a firewall. It has to happen in the messy, frustrating, and all-too-human space between us.