When you research every candidate running for office in Toronto, Mississauga, and Brampton, after a while, they can start to all blend together in your head. It’s not just that there are more than 600 candidates across the three cities. It’s that many of their campaigns do, in fact, look identical.

In their flyers, their headshots are in the same place. The fonts and colours match, right down to the slight tilt of the text on the page. In the background, a fuzzy, not-quite-illustrated image of a nondescript suburban park in bloom, or of Toronto’s waterfront skyline. The same five vague talking points. If you’ve checked the social media pages of even a handful of candidates running for mayor or to be your councillor, you’ve probably seen a version of this poster. A person didn’t design this, not in any meaningful way—generative AI did.

For the past two months, I’ve had the responsibility of overseeing The Local’s election service journalism tool, Candidate Tracker, for three of the biggest cities in Ontario. That has meant either personally looking at or editing someone else’s reporting on every single candidate running for mayor, councillor, or school trustee. I’ve done this twice before for Toronto, during the 2022 general election and 2023 mayoral by-election, and each time, it’s allowed me a bird’s-eye view of the election. And this time, it looks a lot different than it has before.

The 2026 municipal election is the city’s first AI election: the first campaign period during which generative AI is available for mass uptake and delivers results that can land somewhere on the spectrum from passable to good. The natural fear, in this technological paradigm, is of misinformation. And while that concern is more than fair, what I’ve observed instead is a more oblique threat: noise.

It is extraordinarily easy to run for a municipal election in Toronto. Just 25 people need to endorse your nomination, a number set by the province’s Municipal Elections Act and applicable anywhere from Toronto to Latchford, Ontario’s smallest town. As a result, every time there’s an election in this city, I’ve noticed perennial candidates who sign up and shell out the $100 it takes to run, only to not meaningfully campaign at all. No websites, no social media presence, no endorsements.

I always figured these folks just get a thrill out of seeing their name on the ballot on election day. But this year, we’re seeing a lot fewer nonexistent campaigns and a lot more hyper-generic, AI-generated ones instead. It’s turned deciding who to vote for into picking a radio signal out of static—and voters today will have to work harder to tell the serious candidates from the others.

Once we first spotted the AI, it was impossible not to notice more and more. One candidate published a generated video of himself next to likenesses of mayoral incumbent Olivia Chow, and frontrunners Brad Bradford and Chris Alexander (this was early in the race) competing for the job of mayor on a gameshow. Others ran their headshots through ChatGPT, apparent to us because it was in the very file names on their websites. Candidates have sent me lengthy platforms copied whole cloth out of their chats with Gemini and ChatGPT. In a move that is baffling but almost endearingly sincere, one candidate with a background in music therapy centred her entire campaign around AI-generated videos of her singing uplifting songs about fixing Toronto (most of the fixes were also generic; not much rhymes with “mixed-use zoning”).

I even found I was able to recreate a version of the AI campaign flyer I’ve been seeing everywhere by asking ChatGPT—reluctantly, I must preface, and for the first time ever—to “generate me a campaign flyer for the 2026 Toronto municipal election,” and then asking for the resulting flyer to be “a little more fun.” Without my prompting, the flyer included perfunctory platform points I recognized from other campaigns: “more affordable housing, better transit, a cleaner, greener city, safer neighbourhoods, a more inclusive Toronto.” Above that, “A city that works for everyone,” underlined. Even the little graphics accompanying each point, meant to look like hand-drawn doodles, were identical to those I’d seen elsewhere.

To a lay person, these goals make sense, and likely reflect at least some, if not all, of their values. And we live at a time when news consumption rates, trust in media, and voter turnout are flagging. Everyone is competing against social media algorithms for attention, and there’s little glamour in drilled-down policy proposals—like installing crosswalks, negotiating waste management contracts, or arguing for a commercial parking levy—on the campaign trail. Amid all this, I’ve been asking myself: will enough voters look past the buzzwords and know to push, to ask candidates, “How would you achieve this?”

There’s an argument to be made that in some sense, AI has democratized the election process by eliminating the need for expensive marketing teams. I spoke to one candidate, Ketan Purohit, who is running for Toronto District School Board (TDSB) in the board’s new Ward 12. His AI use was easy to spot if you know what to look for: in a generated campaign video, he’s walking through a school hallway, but the locker doors in the background phase in and out of existence; in another scene, an incongruously complex mathematical equation on a whiteboard changes with every shot. But to the untrained eye, this might not be as evident as AI—we’ve come a long way from the early videos of a warped, otherworldly Will Smith eating spaghetti.

Purohit tells me he wanted to use the videos to reach the people he was speaking to  about the issues that mattered to them the most. “Going door to door campaigning, I came across many common questions,” he says. Seniors would share particular perspectives with him, as would parents and young people. Making AI videos catered to each demographic’s concerns felt like a way to connect with those voter bases.

To Purohit, AI feels like an inevitability, and he argues the TDSB needs to adapt its approach to teaching about it, or risk the “new world” leaving kids behind. “We all know that the new generation has already started using it,” he says. It’s a tool. “The next question is how well we use it.”

One of the things that struck me about Purohit’s videos is that his AI-generated self didn’t have the same accent he has in real life (when I ask him about this, he tells me this wasn’t intentional, but a byproduct of the limitations of Google’s AI, Gemini). I’ve seen similar changes with other candidates too: airbrushed people with fuller heads of hair, their dark circles erased, putting themselves in officious suits and against grand backdrops in an endeavour to make themselves more beautiful, more electable. It’s a reminder that generative AI is a product of the society around it.

Toronto Catholic District School Board trustee Maria Rizzo, who is running for re-election and has been in office since 2003, tells me bluntly that ChatGPT tried to make her appear younger the first time she used it. “I would give [it] a picture of me, and then it would generate me like a supermodel,” she laughs. “And I’d go, ‘No, I’m fatter than that!’”

Another time, she says, she asked ChatGPT to generate a short description of her. It described her as good at listening to people. “Not really,” she tells me. “You know, I’m deaf.”

Rizzo’s Instagram features AI-generated posters and videos of her smiling, laying a hand over her heart, made through ChatGPT and TikTok. She has no graphics or communications team for her re-election campaign, she says. This summer has been her first foray into AI use, and she’s not sure she’ll continue: she feels that there’s room for responsible AI use, but she also noticed mistakes in it when she asked ChatGPT for school addresses and bell times. And of course, she recognizes the danger  it poses in an educational setting, for students and for teachers. “There’s no accuracy unless there’s a human factor in it,” she says. (When I ask her how she squares her use with the Pope’s cautious teachings about AI, she first says she’s not too certain. After our call, she texts me to say she asked ChatGPT what the Pope said. It’s quite a moving encyclical, I respond, finding myself quoting the Pope: “Ours is the pressing duty to remain profoundly human.” As it turns out, she agrees with him.)

This election isn’t the first time AI has helped shape municipal politics in Toronto, and it’s certainly not the worst. Last spring, an organization representing critics of the Bathurst Street bus lanes created videos of AI-generated “residents” arguing against the RapidTO transit development. “Anti-woke” advocacy group IntegrityTO and right-wing two-time election candidate Anthony Furey have also used generative AI to fabricate photos of homelessness and drug use in the city. (Furey’s team, memorably, got dinged at the time for letting a generated image of a woman with three arms make it to their platform document.)  This May, after the start of the election period, IntegrityTO posted a deepfake video of Toronto mayor Olivia Chow announcing the weekend closure of the Don Valley Parkway for an absurd set of reasons, including sports games and Mothers Day (the DVP was in fact closed for maintenance). Some who posted comments on the video appeared to believe it was real.