Unless you’re the sort of person who has copies of Brill’s Content moldering in your basement and uses a still of Robert Redford in All the President’s Men or Holly Hunter in Broadcast News as your X avatar (and you’re still on X, come to think of it), you’ve probably never heard of Margaux Blanchard.
A couple of weeks ago, Charlotte Tobitt of the UK-based Press Gazette broke the news that a freelancer using the byline Margaux Blanchard placed dozens of articles in outlets like Wired, SFGate, and Business Insider. These stories were full of unverifiable facts, riddled with persons and places that appear not to exist, and were, almost without a doubt, composed by AI.
Since Tobitt’s story ran, many of the publications retracted and appended editor’s notes onto Blanchard’s stories. This week, The Washington Post ran a follow-up reporting that Business Insider’s internal investigation has cast doubt on several other bylines including Nate Giovanni, Tim Stevensen, and Onyeka Nwelue. The last name is apparently attached to a real person who has been accused of fabricating his credentials in the past.
The internet being a hall of cracked mirrors, someone has already set up a Margaux Blanchard account on X and taken credit for the entire hoax, claiming “a major media client approached me with a unique opportunity. They wanted actionable data: could a fully autonomous AI system produce credible news stories of sufficient quality that they could be sold to top-tier outlets?” The person who wrote (or prompted) that word salad was quickly revealed to be a pseudonym for someone who’d (falsely) claimed credit earlier this year for creating a highly-streamed AI music group.
Whatever.
Here’s the thing with a fraud like this: it muddies the water in such a way that everything that follows it seems a little faker. I’m sure if I look hard enough, there are people questioning if the Press Gazette is a real publication and if Charlotte Tobitt exists. In a world where an AI-created influencer got a bunch of press for “attending” Wimbledon and a chatbot programmed to emulate reality show stars is flirting with people to death, who’s to even say what’s real anymore?
Most digital-first publications are struggling. They’re publishing as much as they can as quickly as they can all day, every day. There are many reasons they’re doing this, almost none of them related to journalism.
I’m less interested in the Blanchard hoaxer(s) motivation in selling these fugazi stories than I am the publications’ culpability in putting them in front of readers. The person or persons who cooked up the Blanchard posts — or, more likely, prompted ChatGPT — understood a major problem in the media industry and exploited it with ease: Desperation.
Most digital-first publications are struggling. They’re publishing as much as they can as quickly as they can all day, every day. There are many reasons they’re doing this, almost none of them related to journalism: to keep up with the neverending scroll of social media, to increase their traffic and get more impressions for ads, to position themselves at the top of Google searches, to build a “long tail” of content that is always findable. Mostly, these publications feel compelled to throw everything at the wall in the desperate hope that something, anything, will stick.
Worse still, they’re doing this with fewer employees thanks to near constant industrywide layoffs. That means that by the time a piece of “journalism” hits your feed, it’s been seen by fewer fact checkers, fewer copy editors, and fewer editors in general asking themselves and one another if it passes the “smell test,” the journalistic equivalent of separating fact from fiction.
The Blanchard hoaxer(s) merely did what hundreds of other freelance content creators are doing all the time: playing the numbers. Writers are throwing huge handfuls of pitches at editors until one takes. A quick search for “how many pieces does business insider publish a day” brought me to a cascade of links like “How to Become a Contributor to Business Insider” and “Business Insider Published My Story in Less Than a Week” full of detailed instructions for how to get ideas accepted and whom to pitch.
Add ChatGPT into this pitches’ brew and you get a content machine that can analyze what works, synthesize the how-to Medium posts, and write the story in minutes.
As an editor, I’ve received dozens of these pitches over the years. They’re never fabulous, usually passable, and in a moment of weakness, totally assignable. At a previous job, one writer sent me so many of these bland ideas that I finally did some clicking around and discovered that the writer, who looked to be in their 30s, had published over 28,000 stories. Back of the envelope math, on, let’s call it a decade-long career, has that person writing around 3,000 stories a year, which seems like a lot for anyone to do if they like to sleep once in a while. (Back in 2006, the media columnist for The New York Observer positively marveled at the fact that a young Sewell Chan had written 422 stories for The New York Times in one year, a positively leisurely pace of work compared to most 28-year-old journalists working today.)
Publications are also desperate to replicate their own successes. If a story does well, editors are quick to assign something similar hoping it, too, will succeed. It’s worth noting that one of Blanchard’s stories ran on SFGate and was about Disney adults, a topic many publications can’t seem to get enough of. The headline of that piece (which was taken down and replaced by an editor’s note) was “‘Toxicity and negativity’: When Disneyland goes from a hobby to an obsession.”
That’s definitely clickable, as was, I assume, the 2021 Gate headline “‘They start calling you Hitler’: Why Disneyland has some of the most toxic fans on the internet,” or “The rude guest habit that's ruining Disneyland rides for everyone else,” which ran earlier this summer. A savvy writer — or a ChatGPT-enabled fraudster — need only look at the stories that do well online to come up with dozens of ideas sure to appeal to a busy, numbers-fixated editor doing the work of the six colleagues their bosses just laid off.
News organizations’ desperation to keep up with the demands of algorithms and search engines while hitting ever higher traffic goals has made them ripe for exploitation by bad actors. It sucks to be taken advantage of; it’s even worse when you make it so easy.
If a story does well, editors are quick to assign something similar hoping it, too, will succeed.
It’s not just high volume, traffic-hungry websites that occasionally get duped. Before our metrics-obsessed era of journalism, publications were also desperate: for readers, ad dollars, and buzz. This opened them up to all kinds of scandals that journalism nerds study the same way doctors are drawn to medical mishaps. These worst case scenarios can remind us what is “normal” and give us a shuddery, vicarious thrill.
In 1976, New York published Tribal Rights of the New Saturday Night, a dispatch from an outer-borough disco that formed the basis of the movie Saturday Night Fever (1977) even though the characters and setting were made up by its author Nik Cohn. (New York’s slipperiness was so well-known that a character in Whit Stillman’s 1990 movie Metropolitan tells a lie and justifies it by calling it “a composite, like New York magazine does.”)
In the 1990s, The New Republic published a series of attention-grabbing yet impossible to verify pieces by Stephen Glass, a case of journalistic malpractice that inspired Shattered Glass (2003).
In the early 2000s, The New York Times had Jayson Blair and Slate had its monkeyfishing and fake auto executive diary. The Village Voice had to retract a cover story about wannabe pickup artists that was largely bogus and led to the firing of the paper’s interim editor.
More recently, The Atlantic attached an extensive note to a piece stating its editors could not “attest to the trustworthiness and credibility of the author,” who had come under criticism for plagiarism earlier in her career at The New Republic.
Even The New Yorker, the best magazine in the world (“probably the best magazine that ever was”) can get taken in despite its punctilious, bordering on neurotic fact checking. In November 2000, the magazine ran a piece called My Fake Job that turned out to be partially fabricated by its author, a Late Show with David Letterman writer named Rodney Rothman.
I thought about Rothman immediately when I read Tobitt’s report about Margaux Blanchard. I remembered how baffled everyone was at the time by why — and how — a writer would trick The New Yorker into publishing a bunch of lies. At the time, Rothman dodged interviews, letting his manager speak for him. (It didn’t hurt him, either: He’s had a successful career in screenwriting and shared a Best Animated Feature Oscar for Spider-Man: Into the Spider-Verse in 2018.)
Maybe Rothman perpetrated his hoax for fun, just to see if he could get away with it. Maybe that’s why Margaux Blanchard and those other fauxlancers did it, too: the lulz.
Here’s the thing: Journalism is at an existential moment. Readers won’t pay for it; tech platforms undercut it; politicians at the highest level are labelling it “fake” and calling its practitioners “the enemy of the American people”; and AI companies are simultaneously hoovering it up to train LLMs and spitting it out as slop, making the work of real journalists less valuable and their lives less livable.
This isn’t the time for jokes, Margaux. Besides, we’ve heard this one before, and it wasn’t funny then.
