When several articles were published last week on Reviewed, a USA Today website that recommends products, something seemed amiss. No one at Reviewed knew the bylines on the pieces.
Reviewed writers and editors began looking up the names, but struggled to find evidence—such as a LinkedIn account—that the people existed. The quality of the items was also questionable. That’s when they started asking themselves: Did artificial intelligence write these articles?
Gannett, the parent company of USA Today, says no AI was used. About 40 people at Reviewed say yes.
Some of the articles in question were reviewed by artificial intelligence recognition programs, which repeatedly found that some of them were not written by humans, a representative of the union that represents Reviewed employees said in an email Friday.
One of these programs, Winston AI, found that three articles had “zero percent human value,” indicating that they were most likely not written by a human The Union. Another had a human value of 1 percent.
One of the articles with a zero percent human score was a recommendation for the best portable trampoline.
“Looking for the best portable trampoline can be daunting,” said the Review said. “Fortunately, this buying guide covers all the essential factors to consider when shopping. Regular use of a trampoline can help improve balance, coordination and mobility.”
According to Winston AI, “It is very likely that an AI tool was used to generate text.”
According to Gannett, that is not the case.
Lark-Marie Antón, a spokeswoman for the company, said in a statement Friday that the articles in question were “created by external freelancers hired by a marketing agency partner, not AI”
However, Ms Antón acknowledged that the reviews had not been properly identified as being written by third parties.
“The pages were provided without accurate affiliate disclaimers and did not meet our editorial standards,” she said, adding that updates to the articles had been published.
Others were demolished after a riot by several workers at Reviewed.
Asked about the articles that Reviewed staff ran through artificial intelligence recognition programs, Ms. Antón said the finding that they were not written by humans was “baseless.”
Reviewed authors and editors are demanding that all articles in question be retracted and that the company apologize for using a third party to do the work it could have done.
“We have been told in no uncertain terms that this is not going to happen,” Garrett Steele, search engine optimization editor at Reviewed, said Friday.
According to a union representing Reviewed employees, the third-party company was AdVon Commerce. The company did not immediately respond to a request for comment Friday afternoon.
As artificial intelligence has become easier to use in recent years, some companies and news organizations have been experimenting with the technology to create content. This has led some authors to worry that their work could be replaced by artificial intelligence.
The Reviewed Union said on X on Thursday that Gannett “will put profits before workers’ rights or journalistic integrity, so we are organized to fight against this attack on unions and the public trust.”
“If AI increases productivity, we demand a fair share and no threat to our jobs,” the union said. “Workers deserve to share in the benefits of new technology, not the risk of being replaced.”
This was announced by the NewsGuild of New York, which represents the Reviewed union on X that verified union members “will NEVER be replaced by AI”
The NewsGuild also said on X that the articles were a “blatant attempt by Gannett to bust the union by threatening reporters with job loss” after members of the Reviewed union staged a two-day strike this month to protest a new contract.
Ms. Antón said Gannett’s claim of union busting was “patently false.”
She added: “Our leadership is focused on investing in our newsrooms and monetizing our content while continuing to negotiate fairly and in good faith.”