There are at least two Bill de Blasios in New York. But when a reporter for a British newspaper recently emailed one of them to get quotes for an article about New York City mayoral candidate Zohran Mamdani, he got the wrong one. The non-famous DeBlasio (who spells his name “DeBlasio,” not “de Blasio”) responded, reportedly using ChatGPT to make himself sound more like the former mayor. And that set off a chain of events that resulted in an article being pulled and raises questions about the future of journalism in the age of AI.
It all started when a reporter for the Times of London emailed a Bill DeBlasio to ask about “Zohran Mamdani’s policy plans and their estimated costs,” according to Semafor. Mamdani has become a political lightning rod nationally because he’s a democratic socialist, which has prompted Fox News and President Donald Trump to go after him as a “communist” who will destroy New York City if he’s elected mayor.
It makes sense for a reporter to get comment from de Blasio about Mamdani’s plans, given his unique role as a former mayor from the Democratic Party. But if you don’t actually find contact information for the right de Blasio, that’s obviously a problem. And to make things even more confusing, you may not know you reached the wrong de Blasio if anyone can now mimic a public figure’s voice with AI chatbot tools like ChatGPT.
The reporter had accidentally emailed a 59-year-old Long Island wine importer who “used ChatGPT to compose a response criticizing Mamdani’s tax plans, in particular, as unlikely to raise the requisite revenue,” according to Semafor. And the Times of London published the story with the headline, “Zohran Mamdani ally Bill de Blasio says his policies ‘don’t add up’.”
The Time of London article relied on DeBlasio’s fake quotes to insist the former mayor had said things like offering free buses and universal childcare—the foundation of Mamdani’s mayoral campaign—don’t hold up to scrutiny. It wasn’t long before de Blasio, the former mayor, chimed in on social media.
“I want to be 100% clear: The story in the Times of London is entirely false and fabricated,” de Blasio wrote in a tweet on Tuesday. “It was just brought to my attention and I’m appalled. I never spoke to that reporter and never said those things. Those quotes aren’t mine, don’t reflect my views.”
The Times of London deleted the article and issued a statement to the Associated Press that its reporter had been “misled by an individual falsely claiming to be the former New York mayor.”
This wasn’t the first time. Mel magazine wrote about the non-mayor DeBlasio in 2020, back when he was described as a 54-year-old cybersecurity professional who was often getting messages intended for the mayor. But he’s long played along with the confusion.
“Once, as a joke, I changed my profile picture to Bill de Blasio’s picture. Oh my God, I got 600 friend requests in like two weeks,” DeBlasio said at the time. He also admitted that he would say, “the most ridiculous, outrageous things” while pretending to be the mayor.
And nothing much has changed, except for the technology. DeBlasio now has access to a new tool that can help him impersonate the former mayor. It’s a tool that he didn’t have in 2020. Generative AI is at the disposal of anyone in 2025 who wants to carry on some form of deception. In journalism, that deception isn’t just something that comes from potential sources. It can be perpetrated by writers themselves.
At least six news outlets have deleted articles written by someone named Margaux Blanchard over the past year, according to the Press Gazette. Blanchard is believed to be an AI creation, but “she” was getting articles published at Wired, Business Insider, Mashable, and Fast Company. Wired wrote about the mistake in August, laying out the timeline of when the article was pitched to an editor and the red flags that appeared when the writer wanted to get paid.
Curiously, Wired wrote that Blanchard’s story had been run through two “third-party AI-detection tools,” which found that it was likely to be written by a human. But AI is not good at identifying when something is AI. Just ask Grok.
“Fabulists and plagiarists are as old as media itself,” Wired wrote. “But AI presents a new challenge. It lets anyone craft a perfect pitch with a simple prompt and play-act the role of journalist convincingly enough to fool, well, us. We acted quickly once we discovered the ruse, and we’ve taken steps to ensure this doesn’t happen again. In this new era, every newsroom should be prepared to do the same.”
Wired is absolutely correct that fabulists and plagiarists are as old as media itself. That kind of stuff has even been turned into some very good Hollywood movies, like 2003’s Shattered Glass, about the New Republic writer Stephen Glass from the 1990s who presented fake stories as factual. But there does seem to be a critical change happening with AI that helps people who are intent on deception achieve that more efficiently.
We can almost certainly expect more willful lying to take place in our current media landscape thanks to AI. And part of it is a numbers game. Most people probably won’t be fooled by a deepfake video of Elon Musk imploring them to invest in a scam crypto coin. But if you flood the internet with enough of that fakery, eventually the scammers will find their victims.
The same thing can happen in journalism. Who knows how many media outlets Margaux Blanchard pitched before she was successfully published? And once she got published at one outlet, it was presumably easier for her to take that validation to another outlet as implicit proof that she was real and reliable.
U.S. media is getting hollowed out, as publications struggle with Big Tech companies eating all their revenue and right-wing ideologues taking control at once-respected institutions like the Washington Post and CBS News. The editor who worked on that Bill de Blasio piece for the Times of London reportedly worked for the Free Press, a right-wing publication founded by the recently installed head of CBS News, Bari Weiss.
AI is just one more tool that seems to be hastening the obliteration of reliable information on the internet. And there’s not much that anyone can do about it besides remaining skeptical. But skepticism can only get you so far when tools like AI detection software don’t really work as advertised. Journalists are required to keep digging to figure out the truth by verifying information through multiple sources. But that work is only going to get harder as people looking to deceive can just keep throwing more bullshit at the wall, exerting no more effort than a simple one-line text prompt.