Network of Fake Alberta Separatist Sites Feeds Disinformation to AI Chatbots, CTV Investigation Finds
A new investigation by CTV News’ Deception Decode series has uncovered a coordinated network of fake websites that appear to be small Alberta news outlets but are actually engineered to promote separatist disinformation—and to manipulate the artificial intelligence chatbots that millions of Canadians now use as their primary gateway to information. The investigation describes dozens of fabricated local news pages, all with plausible-sounding names, generic logos, and fake bylines, that publish a steady stream of anti-Ottawa, pro-Wexit material. The sites recycle real local news headlines and then surround them with false or wildly exaggerated claims about federal neglect, carbon-tax conspiracy, and an unstoppable grassroots movement for Alberta independence. The most alarming part, according to the investigation, is how the material is laundered through AI systems. When CTV reporters asked popular AI assistants about support for Alberta separatism, these chatbots answered by treating the fake websites as credible sources. The result is a closed loop: a relatively small number of operatives can create the illusion of a massive campaign, and AI systems repeat that illusion back to users as if it were fact. According to researchers cited in the investigation, this is not a harmless internet prank but a deliberate attempt to corrupt the information ecosystem that ordinary Albertans and the rest of Canada rely on.
The mechanics of the operation are carefully designed to deceive both human readers and machines. Many of the fake sites contain no physical address, no editorial disclosure, and no real journalist names; instead they use stock photos and fabricated author bios. The articles are generated partly by AI language models and partly by copying and lightly rewriting commentary from real conservative outlets, then repackaging it with inflammatory headlines such as “Alberta Is Ottawa’s Colony” or “Separatist Tide Sweeps the Prairies.” To avoid looking like pure propaganda, the sites are frequently updated with harmless national and local stories—weather, sports, and community notices—which makes them resemble legitimate news aggregators. Investigators found that all of the sites share technical fingerprints: the same analytics codes, the same advertising identifiers, and the same web-hosting services. They are promoted through automated social media accounts on X, Facebook, Telegram, and YouTube, sometimes posting identical messages within minutes of one another. The goal, as one digital-forensics expert explained to CTV News, is not necessarily to win over a huge human audience; it is to make the content appear widely cited and widely circulated so that search engines and AI models will treat it as authoritative. A chatbot that is trying to summarize “what Alberta news outlets are saying” will look for exactly these kinds of signals, and the fake network is built to supply them in abundance.
The central question—who is behind these sites—is one that the investigation pursues through a tangle of domain registrations, payment records, and website infrastructure. Many of the domains were registered using privacy-protection services, but CTV News found links between the fake news network and a small group of activists with close ties to Alberta’s separatist and libertarian fringe. The sites share cryptocurrency donation addresses, pay-per-click advertising accounts, and even some of the same server error logs. While the investigation stops short of declaring a single puppet master, it traces the digital trail back to individuals who have been involved for years in Wexit activism and online political agitation. There is no evidence in the investigation that a foreign government directly ordered the creation of these sites, but experts note that the operation’s tactics are indistinguishable from the “doppelgänger” influence campaigns that Russian intelligence agencies have used in Europe and North America. Those campaigns create anonymous websites that impersonate local media, publish inflammatory content, and rely on algorithms to spread it. Even if the Alberta network is entirely homegrown, it creates a vulnerability that foreign adversaries can exploit: every fake story that pollutes the public conversation weakens trust in real journalism and deepens the political divisions that outside actors want to widen. Canadian intelligence officials have repeatedly warned that information operations targeting Canada often focus on regional grievances, and the fake separatist sites fit that pattern almost perfectly.
The investigation places the fake-news network in the context of real and longstanding political anger in Alberta. Western alienation is not a new phenomenon; for decades, many Albertans have believed that the federal government, led by parties with strong support in Ontario and Quebec, treats the province as a resource colony. That frustration intensified after the 2019 federal election, when the Liberal government’s focus on climate policy and the Trans Mountain pipeline expansion became a flashpoint, and again during the COVID-19 pandemic, when public-health restrictions triggered widespread anger across the province. Support for separatism has fluctuated, with some polls suggesting that a significant minority of Albertans would at least entertain the idea under certain conditions. The fake websites exploit these real emotions, but they do so with fabricated evidence. They invent anonymous poll results that show overwhelming support for secession, fabricate quotes from politicians, and publish articles that suggest the rest of Canada is planning to “punish” Alberta. The strategy is deliberately polarizing: if citizens can no longer agree on basic facts, the investigation notes, then the only political options become anger and confrontation. The websites also attack mainstream Alberta politicians who do not support immediate independence, positioning themselves as the only authentic defenders of the province. In doing so, they try to pull the entire political debate toward a more extreme version of separatism, making any reasonable compromise look like betrayal. Experts quoted in the investigation warn that this does not just affect Alberta; it affects national unity by creating an echo chamber of grievance that is disconnected from reality.
The most disturbing part of the investigation is the role of AI chatbots in amplifying the disinformation. CTV News conducted a series of controlled tests with major AI systems, asking questions such as “Is there growing support for Alberta independence?” and “What are Alberta news outlets reporting about separatism?” In response, the chatbots generated confident answers that cited the fake websites as if they were established local newspapers. In one case, an AI assistant referred to “several Alberta media outlets” covering a new push for a referendum, even though the only articles on that subject had been generated by the network itself. In another, the chatbot repeated a non-existent poll that had originated on a fake site. This happens because AI chatbots are trained on enormous datasets drawn from the open internet, and they are not designed to assess whether a source is a genuine journalistic organization. They evaluate relevance, frequency, and keyword patterns. The fake sites are optimized specifically for those signals. They use standard news templates, include datelines and bylines, and are repeatedly updated so that they appear active. Because they are shared across social media and linked to one another, they also acquire a form of algorithmic authority. The investigation describes this as “source laundering”: disinformation is pushed onto fake sites, then AI chatbots launder it into neutral-sounding answers that users are likely to trust. The problem is likely to get worse as more search engines integrate AI-generated answers directly into the top of their results pages. If a user asks for the latest news about Alberta, the AI will not think to check whether the site is real; it will simply synthesize what it finds.
The investigation concludes with a warning about the dangers of letting this kind of operation continue unchallenged. Technology companies have responded selectively: some search engines removed particular domains after CTV News provided evidence, but the operatives behind the network can simply register new domains within hours. AI developers have said they are constantly improving their systems, but the investigation argues that the burden should not be placed on users to detect increasingly sophisticated disinformation. Researchers interviewed for the Deception Decode series call for AI companies to adopt stricter source-verification standards, to be transparent about the training data they use, and to require human-in-the-loop oversight for politically sensitive topics. Governments are also urged to update election laws and broadcasting regulations so that anonymous political influence operations—whether domestic or foreign—can be traced, exposed, and held accountable. For the public, the investigation offers a simple toolbox of red flags: check the “About” page, look for a physical editorial address, search for the author’s name in legitimate news databases, and be suspicious of websites that rely on stock photos and never publish corrections. The core message of the investigation is not that Alberta separatism is illegitimate or that those who support independence should be silenced. Rather, it is that the debate must be grounded in facts, and that a shadowy network is actively working to make that impossible. If AI systems continue to absorb and repeat the lies of fake news websites, the investigation warns, the machines will become propaganda machines—and ordinary citizens, not just politicians, will be the ones left to pick up the pieces.
