A true sign of the vast shift in the media landscape, a well-established newspaper was deceived into publishing an “opinion piece” that was generated on ChatGPT.
The Irish Times, which has been in business for 164 years, unwittingly ran an AI-written article that came complete with a fake author’s photo in its Opinion section last Thursday. The oversight was the result of what the paper deemed as a “deliberate and coordinated deception” by a third party.
The faux thinkpiece, entitled Irish women’s obsession with fake tanis problematic, condemned Irish women for darkening their skin unnaturally, a beauty ritual that the “journalist” had claimed to be a mockery of people with real, dark complexions.
This is interesting. Was it a hoax article in Irish Times? Was it a deliberate scam, written using AI? Does the person exist? Is it indicative of what we may see more of? pic.twitter.com/hlNDFZdLZf
— Aidan O’Sullivan (@AidanOSulliva15) May 12, 2023
The article was “written by someone purporting to be a young immigrant woman in Ireland,” detailed the publication’s editor Ruadhán Mac Cormaic in a statement on Sunday. “It made an argument that has been aired in other countries but related it to the Irish context.”
Per the Guardian, the write-up—submitted by a troll—supposedly became the newspaper’s second-most-read article ever, and its contents even sparked a discussion on the radio.
Apologizing for publishing the hoax article, Mac Cormaic added: “It was a breach of the trust between the Irish Times and its readers, and we are genuinely sorry.”
Mac Cormaic reflected that the event “highlighted a gap in our pre-publication procedures” and “also underlined one of the challenges raised by generative AI for news organizations.” He promised that the newspaper would work to make its approval process “more robust.”
The “writer” behind the piece had assumed the name of Adriana Acosta-Cortez, and was described as a 29-year-old Ecuadorian immigrant and health worker living in north Dublin. The character’s photo depicted “her” with light blue hair.
Might be wrong about this but something seems really off about the author photo that ran with this Irish Times fake tan article. @keyes and I ran it through a few AI checkers and results are.... https://t.co/3KYQKleYZ8pic.twitter.com/ZsNqtuLIFN
The newspaper’s editor said “Acosta-Cortez” had gone back and forth with the editorial desk for several days to make corrections, as well as provided personal anecdotes and links for references.
Behind the cotton-candy blue tresses, however, was a different face. The prankster, who has remained anonymous, admitted to the move using a fake Twitter account with the same bogus name and picture.
— Adriana Acosta-Cortez (@ecuadorian_adri) May 14, 2023
They later told the Guardian that they just wanted to “give [their] friends a laugh” and “stir the s**t” amid the wave of identity politics.
The author, who claimed to be a student, disclosed that they utilized ChatGPT to produce about 80% of the opinion piece.
They also used DALL-E 2 to generate a faux profile photo using the prompt, “female, overweight, blue hair, business casual clothing, smug expression,” an image the author had perceived as the ultimate description of a “woke” journalist.
The divisive article was deleted hours after its publication.