Karen Mazurkewich is vice-president of stakeholder relations and communications at Toronto Pearson. This opinion piece originally appeared at Fast Company.
Media relations has always been a fast-moving puzzle —part science, part art, all delivered in a sprint on someone else’s deadline. What are the facts? Who is the reporter? How do we get errors corrected?
I deal with such questions daily in my role as an airport-communications executive. But as the media industry becomes more digital, the challenges and pace only increase. Today’s “reporter” might be a clickbait farm, a social-media pseudonym or a casino looking to boost SEO for brand awareness—all cases I’ve seen recently.
But the changes artificial intelligence creates are on a different scale altogether, opening challenging fronts in the message wars. Algorithmic content operates at a speed humans can’t keep up with. At least half of all web articles are already AI-generated, according to a May 2026 study by digital marketing firm Graphite.
In this brave new world, the writers are bots, the errors are uncorrectable, and AI’s “Trust me, bro” interface separates readers from the original sources of information they’re consuming online.
To be clear, I’m no resister. AI will boost efficiency for communicators and others in ways we’ve only begun to understand. But as these tools become increasingly standard on our devices and in our feeds, we need to watch out for misuse.
The bot journalist reports
One bad use case is bot-written content posing as journalism. One piece I saw recently showed all the hallmarks. “Toronto Pearson disruptions ripple across North America,” it said, referring to 95 flight delays and 14 cancellations on April 9 that created increased disruption over several consecutive days.
This piece was puzzling for several reasons.
First, the “elevated disruption” never happened. Our airport recorded completely normal operations on April 9—indeed, slightly better-than-average outbound delay and cancellation rates that day and three days prior.
You would have had to go back nearly two weeks to find a single instance where our airlines’ departure performance was anything but standard—a day in March when thunderstorms affected travel across North America’s eastern seaboard.
This site didn’t contact us, so I don’t know what figures or logic went into “elevated disruption.” This is the stuff we help human reporters get right.
Second, the 1,200-word piece was classic AI bot style: grammatically flawless, confident in tone, but at its crux, mostly generic background. The site regurgitates similar content with unsourced information and bland sentence structures over and over again, each featuring a fresh data point followed by hundreds of words of generic background. The airport pieces are illustrated with AI-generated images that look like photos but aren’t.
Nearly 60(!) pieces were published that day under a single bylined journalist, whose photo also sometimes appears under a different name. In auditing the site, we found this was actually a slow day for the journalist, one of several profiles regularly responsible for 120-plus articles a day. If these were real people working 12-hour shifts, they’d be cranking out 1,200 words every 6 minutes.
The business of clickbait AI news sites
This type of website’s business model is often alluded to in the fine print. It’s not to reliably inform anyone about what’s happening at airports or in other aspects of travel, but to generate search engine hits and clicks on affiliate sales links.
You might reasonably ask what makes this clicky business model so different from the ones used by legitimate news organizations, self-funded journalists, influencers, Substackers, and even clickbait websites. The difference is that those content creators are real people, limited in output. None are capable of hijacking feeds and searches with reams of dubious content.
Human reporters and editors are also, at least in theory, accountable for what they publish. But if a bot makes a mistake, who fixes it? Not the journalist in question, whom we couldn’t reach despite multiple attempts to contact the website. None of the website’s bylined authors has identifiable direct contact details or social accounts, almost certainly because they don’t really exist.
All this effort to analyze and push back against one bad piece, when this website piles them up faster than we can respond.
The battle against bots
Remember the infinite monkey theorem? AI can already write the complete works of Shakespeare. But the real problem isn’t in reproducing classic fiction—it’s that bots are starting to write our stories for us, and disseminate them even if wrong.
That’s why, for anyone responsible for an online brand narrative, it’s important to fight back against bad AI content with as much trustworthy information as you feel comfortable sharing. Because what the bots gain in quantity, humans gain in quality—as long as we insert ourselves into the conversation.
Our team has been analyzing bot and chatbot output to understand where the gaps are, including missing chapters (new information that doesn’t appear in responses), stale loops (outdated perceptions that continue to surface), and lack of context (criticism without nuance).
Once you know which facts and narratives are lacking, use that information to produce and share trustworthy content that people will read, like and share.
We’re baking this philosophy into our strategies for news releases (even if they don’t get covered), storified web content, social posts, and even podcast transcripts. The algorithms prioritize recency and engagement. We’ve learned that you can’t correct the bots except by choosing what to feed them.
Indeed, the same reputational lessons could apply to the rest of us. You might have good reasons for not sharing much detail on LinkedIn and Facebook profiles, but if you want people to understand your story, you need to tell it early and often. Because if you leave too many gaps, you’re outsourcing your identity and narrative to the bots—and good luck getting a correction.