A Local News Site in Canada Was Run by Artificial Intelligence and Published False Information
Imagine opening your news feed and finding a bombshell story about a politician in your city who allegedly resigned after a series of controversial tweets resurfaced online. The story has all the hallmarks of a legitimate scoop: details about the content of the posts, a statement of regret from the politician, and a promise from the party to head in a new direction. Now imagine finding out that absolutely none of it happened. The politician never resigned, never held the position mentioned in the article, and never wrote anything remotely close to what the site described. That is exactly what happened with Surrey Speak, a site based in Surrey, British Columbia, Canada, that presented itself as a legitimate local news outlet but was actually operated almost entirely by artificial intelligence.
The case came to light thanks to an investigation conducted by the Investigative Journalism Foundation (IJF), a Canadian organization dedicated to investigative journalism, which revealed a sophisticated scheme of automated fake content production disguised as community news coverage. Surrey Speak had more than 10,000 followers on Facebook, published dozens of articles per day, and had a visual and editorial appearance convincing enough to fool thousands of people in the Vancouver metropolitan area.
The Fake Story About William Azaroff
The most striking example of content fabrication by Surrey Speak involved Canadian politician William Azaroff. The site published an article claiming that Azaroff had resigned from his position as executive director of a municipal party after controversial tweets resurfaced involving issues of race, Indigenous identity, and political ideology. The article reported that Azaroff had issued a statement of regret and that his party had promised to chart a new course.
The problem is that none of it was true. Azaroff never held the position of executive director of the OneCity party, so he could not have resigned from it. In reality, he had just been nominated as the party’s candidate for mayor in an upcoming election. And as for the supposedly controversial tweets? Azaroff, described as a quiet man who runs a nonprofit housing development company, stated that he never wrote anything resembling what Surrey Speak described.
The smallest bit of fact-checking would show that it wasn’t true, Azaroff told the IJF. His team even considered the possibility that the article was a politically motivated attack, but even that theory seemed unlikely. According to him, if an opponent had done it, they wouldn’t have made such obvious mistakes.
Dimitris Dimitriadis, director of research and development at the New York-based company NewsGuard, offered a more likely explanation. He suspects that the artificial intelligence behind the site detected that Azaroff’s name was trending in searches because of his nomination as a mayoral candidate and, from there, fabricated a completely false story to capitalize on search traffic. If a topic is trending and generating a lot of SEO, these sites use the topic to create content about it, Dimitriadis explained. And that content can be completely nonsensical or even false.
Fictitious Journalists and Industrial-Scale Content Production
One of the most revealing aspects of the investigation was the discovery that the supposed journalists at Surrey Speak simply did not exist. The site listed authors with names like John N., Lucas, Riley Anderson, and Ellie Jones, but none of them corresponded to an identifiable real person. There was no contact information for these writers, no verifiable professional profiles, and no indication that human beings were actually writing or reviewing the published content.
And the productivity of these fictitious authors was superhuman — literally. John N., the supposed author of the fake story about Azaroff, published six other articles on the same day. He and the other ghost writers published a full slate of articles daily, many of them based on real news events that occurred in the Vancouver metropolitan area. The site never disclosed that it used artificial intelligence to create its articles. It was only after the IJF began asking questions that Surrey Speak started labeling some images as AI-generated.
This pattern is typical of the more than 2,600 AI-generated news sites that Dimitriadis has cataloged over the past three years through his work at NewsGuard. According to him, these sites generally work by scraping content from reliable journalistic sources and producing automated summaries. The resulting articles tend to be generic, formulaic, and vague, with few direct quotes, specific references, or credits to the news outlets that originally reported the information.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Fabricated Images of Real Crime Victims
Perhaps the most disturbing aspect of Surrey Speak‘s operation is the use of artificial intelligence-generated images based on real crime victims. An analysis of the site’s source code revealed that at least some of the published images appeared to have been created by ChatGPT.
One particularly serious case involved coverage of a violent robbery that occurred in New Westminster, British Columbia. Surrey Speak published an article about the incident accompanied by an image showing an elderly woman struggling against an attacker wearing a gray hoodie. The robbery actually happened, as confirmed by official sources. However, Suki Sadhre, the real victim’s son, told the IJF that the woman depicted in Surrey Speak‘s image was not his mother. A comparison between a frame from the actual robbery video and the image published by the site showed clear differences, confirming the artificial nature of the photo.
Another case involved coverage of the conviction of Everton Downey, a British Columbia man convicted of murdering his 25-year-old girlfriend, Melissa Blimkie, in 2025. Surrey Speak published an AI-generated photo of Blimkie. The IJF verified the image’s authenticity using two separate AI detection tools, and Dimitriadis performed an additional check using the Hive service used by NewsGuard. All three analyses confirmed the photo was generated by artificial intelligence. A member of Blimkie’s family contacted by the IJF described the use of the image as concerning.
In a more comical but equally telling case, an article about British Columbia’s budget was accompanied by an image that included an inaccurate illustration of a thousand-dollar Canadian bill — an extremely rare denomination that ceased being legal tender more than five years ago.
Factual Errors in Other Articles
The story about Azaroff was far from the only example of serious factual errors. A recent article bylined by John N. falsely claimed that unionized Amazon workers at a warehouse in Delta, British Columbia, had won a significant pay raise through collective bargaining.
In reality, those workers never signed a collective agreement. They did receive a raise, but only because their union successfully argued before a tribunal that Amazon had unfairly excluded them from a wage increase already granted to other non-unionized workers. The difference between the two versions is enormous and has direct implications for how the public understands labor relations in the region.
The Connections to ADGTech and the Network Behind the Site
Surrey Speak did not identify its owners or operators. A phone number and email listed on the site did not respond to the IJF’s inquiries. But the investigation found evidence that appears to link the site to ADGTech Solutions Inc. and its CEO, Anuj Sayal.
An earlier version of the Surrey Speak website listed contact information that included ADGTech’s general Canadian mailbox. It also included a phone number listed by ADGTech. And the site listed an official address in downtown Vancouver that was also used by ADGTech. When Surrey Speak changed its contact information to a different number, that new number was used by ADGScribe, a project of ADGTech itself.
The site was listed as a project of a mysterious company called Network 247, which described itself as Canada’s fastest-growing digital influence network and claimed to reach millions of people in British Columbia monthly through social media pages called I Love Vancouver and I Love Surrey. Network 247 used the same physical address as Surrey Speak and previously stated on its website that it was powered by ADGTech.
The IJF also called Pankaj Sayal, listed as a director of ADGTech in British Columbia alongside Anuj Sayal. Pankaj Sayal confirmed that he was familiar with Surrey Speak and said over the phone that Anuj is the one who handles that.
ADGTech claims on LinkedIn that it employs more than 200 people and has received support from Simon Fraser University’s VentureLabs, described as Vancouver’s leading tech accelerator. The physical address listed by ADGTech and Surrey Speak was actually a coworking space operated by SFU VentureLabs in downtown Vancouver. When the IJF visited the location, a receptionist said that ADGTech did not actually rent space there and that neither Surrey Speak nor Network 247 were affiliated with VentureLabs.
Despite all these apparent connections, Vivek Kumar, who describes himself on LinkedIn as head of business development and the public sector for ADGTech in New Delhi, India, told the IJF that his company does not operate, manage, or exercise editorial control over Surrey Speak. Kumar did not respond to follow-up questions about the numerous apparent links between his company and the site. The IJF contacted Anuj Sayal multiple times by email and phone calls but received no response.
In a profile published in 2025 on the VentureLabs website, Sayal stated that ADGTech had started as a marketing company and was now focused on offering detailed insights for businesses looking to grow online through AI. And he added something that sounds especially ironic in light of the investigation’s findings: At ADGTech, we prioritize ethical AI across all of our products.
The Business Model Behind AI Content Farms
Journalism is not known as a lucrative industry. But with virtually zero operating costs, AI-generated news sites can still generate revenue from online advertising as long as they get enough page views. Dimitriadis explains the logic straightforwardly: the more articles you produce, you basically just need one of them to go viral. You don’t need all of them to go viral. And if you can get one viral article that generates a lot of page views and page impressions, you’re on the path to something self-sustaining and perhaps even profitable.
Surrey Speak was not displaying ads on its page at the time of the investigation, but it listed a marketing email on the site, indicating it was seeking advertisers. Network 247, in turn, also listed a second project called Maple Newswire, whose website domain was also connected to ADGTech. Anuj Sayal and ADGTech recently posted a job listing on LinkedIn for the MaplePulse Network, with a link to Maple Newswire, which was advertised as Canada’s fastest-growing digital influence and civic engagement platform, using proprietary AI-based tools to help brands and public figures trend among communities.
The Consequences for Journalism and Public Trust
Angela Misri, associate professor of journalism at Toronto Metropolitan University, described the content produced by sites like Surrey Speak as AI slop. According to her, these sites are making it even harder for people to separate fact from fiction and, in the process, are eroding the already fragile trust the public places in the media.
People don’t understand that news is not the same thing as just content. And these are just slop content farms, Misri said. This is directly competing with newsrooms for the attention of our supposed audience. We’re going to see a lot more of this. AI slop begets more AI slop.
That last observation from Misri points to one of the most concerning long-term risks. Once false information is on the internet, it can be absorbed and reused by another AI engine, which will treat it as if it were legitimate data without detecting the error. Whether they pulled the information from Reuters or from Facebook, they’re treating it all the same way, Misri warned.
Azaroff himself expressed concern that the fake article published about him could show up in AI-powered searches on engines like Google. It bothers me because it’s so obviously AI slop or just intentional misinformation. The timing seemed very suspicious, he said.
Surrey Speak removed the article about Azaroff from its site on February 23, ten days after the original publication. It also removed the AI-generated photos of Melissa Blimkie and Sadhre’s mother after the IJF inquired about them. The site did not publish any retraction notice or provide a reason for removing the posts.
A Growing Global Problem
Researchers say that Surrey Speak is just one among thousands of AI-generated news sites popping up in Canada and around the world. Dimitriadis, who has cataloged more than 2,600 of these sites over the past three years, said he is particularly concerned about the growth of this trend over the last year. He argues that by competing with real news outlets for limited advertising revenue, these sites will further tighten the financial squeeze on legitimate journalism publications.
Part of the challenge, for me, is trying to figure out how to prevent these sites from gaming the advertising industry and fooling real brands into thinking this is a legitimate place for their ads to be, Dimitriadis said.
For Misri, the risk is even deeper and more philosophical. As AI-generated content takes up more and more space on the internet, she fears that search engines and AI assistants will consume less real journalism and start feeding off each other’s content. The result would be an ocean of slop that drifts further and further from factual reality.
The truth is becoming the minority on the internet. And that’s the scary part, Misri concluded. 😬
What This Case Teaches Us About Information Consumption
The Surrey Speak case is a stark reminder that the appearance of professionalism does not guarantee the truthfulness of content. Sites with clean layouts, journalistic language, and frequent publishing can be nothing more than automated content fabrication operations. The IJF investigation showed that behind a facade of local media committed to real people, real issues, and real change — as the site’s own Facebook slogan stated — there was a system that published lies about real people, fabricated images of crime victims, and showed zero commitment to factual truth.
Artificial intelligence is a powerful tool that is transforming countless industries, including journalism itself. Serious newsrooms around the world are exploring ethical and transparent ways to use AI to assist with tasks like transcription, data analysis, and translation. But what separates responsible use from abuse is the presence of human editorial oversight, an unwavering commitment to fact-checking, and transparency with the public about how and why the technology is being used.
Paying attention to where the news we consume and share comes from is no longer just a good habit. In a landscape where producing convincing misinformation has become absurdly cheap and fast, verifying before sharing has become a responsibility for all of us. 👀
