On March 25, The Atlantic, an American journalistic institution, published an article about suspected AI-generated content in another American journalistic institution, The New York Times. Spoiler: by its fourth paragraph, the article reveals that Kate Gilgan, a writer from Saskatchewan, “did utilize AI as a tool” to write a first-person “Modern Love” column published last November in The Times. Ms. Gilgan said she did not copy and paste AI-generated content into her submission, but rather “used AI as a collaborative editor.”
Did that meet The New York Times’ editorial standards? A Times spokesperson told The Atlantic in an e-mail: “Journalism at The Times is inherently a human endeavor. That will not change. As technology evolves, we are consistently assessing best practices for our newsroom.” And the article remains on the nytimes.com website without an editor’s note. So ... yes?
Less than a week later, The Times added a note to the top of a book review it had published in January, stating: “This review included language and details similar to those in a review of the same book published in The Guardian. We spoke to the author of this piece, a freelancer reviewer, who told us he used an A.I. tool that incorporated material from the Guardian review into his draft, which he failed to identify and remove. His reliance on A.I. and his use of unattributed work by another writer are a clear violation of The Times’s standards.”
Neither of these articles would have met The Globe and Mail’s standards. The Globe’s AI policy, which was updated last fall, specifies that both staff and outside contributors should not use AI to edit or write any part of a story. Although AI detectors exist – and a computer-science professor interviewed by writer Vauhini Vara for The Atlantic used one called Pangram to assess The New York Times’ “Modern Love” column – they are notoriously inconsistent, and so not a useful tool for preventing the publication of undisclosed AI content.
In experimenting with the detectors myself, I have found, as did The Atlantic, that three different AI detectors will provide three different assessments regarding the proportion of an article likely to have been produced by artificial intelligence. “One challenge with AI detection is that the tools involved, much like the models they analyze, are still evolving. Sometimes they flag false positives or fail to catch AI-generated material,” Ms. Vara noted.
The Globe and Mail’s contracts require contributors to departments such as Opinion, First Person and Lives Lived to attest that their work is “original” and created without the use of artificial intelligence. The Atlantic, similarly, “requires contributors to attest to being ‘the sole author’ of their article, and forbids AI-generated writing or imagery without approval and disclosure,” wrote Ms. Vara.
In November, about two weeks after the publication of my column on The Globe’s updated AI policy, I received an e-mail from Nicholas Hune-Brown, executive editor of The Local magazine, and a past Globe contributor. He asked whether The Globe had ever published a writer with the name Victoria Goldiee. In a story pitched to Mr. Hune-Brown, “Victoria” claimed to have written for The Globe. I checked with colleagues. A couple had been pitched, but nothing had ever been greenlit.
Suspicious, in part because “Victoria” seemed to have conducted a number of interviews even before being officially assigned, Mr. Hune-Brown asked the writer whether the extensive quotes she had included in her pitch were from interviews she had conducted, and requested samples of her published work. In his article about the encounter, Mr. Hune-Brown wrote: “Victoria’s stilted email, and a closer read of the original pitch, revealed what should have been clear from the start: with its rote phrasing (‘This story matters because of... It is timely because of... It fits your readership because of…’), it had all the hallmarks of an AI-generated piece of writing.”
I asked him earlier this week how his assigning practices have changed since then. In an e-mail, he said, “the ‘Victoria Goldiee’ saga forced us to make a lot of changes. We finally wrote up an AI policy (which we’ll keep changing as the tech changes: https://thelocal.to/ai-policy/) and amended our contract to explicitly prohibit use of generative AI in crafting Local stories. We also tightened up our fact-checking process, requiring annotated drafts from all writers.”
However, he added, not all of the changes have been positive. “Since the Goldiee saga I haven’t put out a public call for pitches. I’ve been sending pitch calls out to the list I have of potential Local contributors, but I haven’t put anything out to the general public through social media, as I used to. That’s because in 2026, whenever a call for pitches goes public, an editor’s inbox becomes absolutely inundated with AI garbage from around the world. It becomes impossible to wade through all the BS. This is, obviously, a brutal situation for actual human freelancers.”
The Globe and Mail is fortunate to have a large staff of writers who annually renew their pledge to abide by The Globe’s Editorial Code of Conduct. It also has established relationships with a number of trusted freelance contributors, but is always open to pitches from new writers. Writers who don’t already have a relationship with The Globe can access e-mail addresses for assigning editors on the website’s Contact Us page. Assigning editors convey The Globe’s AI policy and take steps to ensure new contributors are who they say they are. That might include having a video call prior to locking in an assignment.
At The Local, Mr. Hune-Brown said, “Our next steps are figuring out how to keep getting pitches from promising young journalists – to keep ourselves open to writers we don’t know, from communities underrepresented in Canadian journalism – without making our lives impossible. That’s a work in progress!”