News agency CNET said Wednesday it has issued corrections to a number of articles, including some it described as “significant,” after using an artificial intelligence-powered tool to help write dozens of stories.
The outlet has since taken a break from using the AI tool to generate stories, CNET editor-in-chief Connie Guglielmo said in an editorial Wednesday.
The revelation comes after CNET was previously publicly called out for quietly using AI to write articles and later for errors. While using AI to automate news stories isn’t new — the Associated Press began doing so nearly a decade ago — the issue has gained renewed attention amid the rise of ChatGPT, a viral new AI chatbot tool that can quickly generate essays , stories and song lyrics in response to user prompts.
Guglielmo said CNET used an “in-house designed AI engine,” not ChatGPT, to help write 77 published stories since November. She said this represented about 1% of the total content published on CNET during the same period, and was done as part of a “test” project for the CNET Money team “to help editors create a set of basic explanations around financial services topics.”
Some headlines from stories written with the AI tool include: “Does a mortgage affect private mortgage insurance?” and “How to close a bank account.”
“Editors first generated the outlines for the stories, then expanded, added and edited the AI drafts before publication,” Guglielmo wrote. “After one of the AI-assisted stories was cited, rightfully so, for factual errors, CNET Money editorial staff did a full audit.”
The result of the audit, she said, was that CNET identified several stories that required correction, “with a small number requiring significant correction.” CNET also identified several other stories with “minor issues such as incomplete company names, transposed numbers, or language that our senior editors deemed vague.”
A correction, which was added at the end of an article titled “What is compound interest?” states that the story initially offered some wildly inaccurate personal finance advice. “An earlier version of this article suggested that a saver would earn $10,300 after a year by depositing $10,000 in a savings account that earns 3% interest compounded annually. The article has been corrected to clarify that the saver would earn $300 on top of the principal of 10,000 dollars, says the correction.
Another correction suggests that the AI tool is plagiarized. “We have replaced sentences that were not entirely original,” according to the correction added to an article on how to close a bank account.
Guglielmo did not say how many of the 77 published stories required corrections, nor did she break down how many required “major” fixes versus more “minor issues.” Guglielmo said the stories that have been corrected include an editor’s note explaining what was changed.
CNET did not immediately respond to CNN’s request for comment.
Despite the problems, Guglielmo left the door open to resume using the AI tool. “We have paused and will resume using the AI tool when we feel confident that the tool and our editorial processes will prevent both human and AI errors,” she said.
Guglielmo also said that CNET has more clearly disclosed to readers which stories were compiled using the AI engine. The outlet took some heat from critics on social media for not making it blatantly clear to the public that “By CNET Money Staff” believed it was written using AI tools. The new byline is just: “By CNET Money.”