AI Embarrassed The New York Times. Will it Embarrass You?
Editors and managers should routinely ask writers about the use of artificial intelligence before publication
Last month, The New York Times warned its freelance writers about generative artificial intelligence after the nation’s second largest newspaper published a book review that copied language from a review in The Guardian.
The Times last month also admitted that its Canada bureau chief wrote a story with an AI-fabricated quote attributed to the leader of the Conservative Party.
The leaders of “The Gray Lady” likely turned pale at the lapses of journalistic ethics.
The episodes point up a common flaw in the AI policies of news outlets as well as corporate communication departments: no requirement to disclose the use of generative AI to editors or managers before publication.
No organization wants to be seen copying another organization’s content or publishing falsehoods, and corporate communication teams also want to avoid these reputation-damaging blunders.
The Times has had worse scandals without AI in its nearly 175-year history. Both AI failures are clearly against the paper’s policies for staff reporters and writers who aren’t employees.
No policy can prevent all such blunders. Even the tech-savvy news organization can’t escape such mistakes.
We’ve offered nine tips on how to talk to writers before and after they’ve written their stories. Now we would add a 10th: Ask whether the writer used artificial intelligence, and how.
‘Dropped in’
A reader (obviously a careful one) alerted the Times that a Jan. 6 book review by novelist Alex Preston “included language and details similar to those in a review of the same book” by the Guardian, according to a Times’ Editors’ Note not published until March 30.
In a statement to the Guardian the next day, Preston said, “I made a serious mistake in using an AI tool on a draft review I had written, and I failed to identify and remove overlapping language from another review that the AI dropped in.”
The Times cut ties with Preston, who as a freelance writer worked on a story-by-story basis. The newspaper said it found no evidence of plagiarism in Preston’s six previous reviews.
Do not enter
In the Times’ May 12 email, the newspaper reminded freelance writers of a brightline rule.
“Freelancers also should not enter any materials intended for use in connection with their work for The Times, or any other copyright-protected material, whether in draft or published form, into gen-AI tools.”
Plain enough. The email was posted by veteran Canadian journalist Karyn Pugliese on her Substack. She also posted the newspaper’s “Newsroom and Opinion Policy on Freelancers’ Use of Generative AI Tools.”
Preston doesn’t explain how “AI dropped in” the Guardian story. Many writers ask AI to summarize background information about a topic.
A fundamental flaw of content generation programs such as ChatGPT, Claude and Gemini is that they vacuum up copyrighted content from across the internet. The output often contains verbatim language from original sources.
The Times’ policy for freelancers acknowledges that it limits the utility of AI. The newspaper suggests using the new technology for “high-level brainstorming without inputting any copyright-protected materials or Times journalism, published or unpublished.”
‘Turncoats’
May began with the Times admitting that Matina Stevis-Gridneff, the chief of the newspaper’s Canada bureau, published on April 14 a made-up quote attributed to a speech given by Pierre Poilievre, the Conservative leader.
In the manufactured quote, Poilievre denounced members of the House of Commons who switched parties to join the Liberal Party.
“Except he didn’t say that,” Pugliese wrote. “Not like that. Not in those words.”
AI summarized Poilievre’s opinion and turned it into an exact quote. In particular, AI added spice by imagining that Poilievre labeled those politicians “turncoats,” which he did not say, as the Times expressly noted in the correction.
“The reporter should have checked the accuracy of what the A.I. tool returned,” the Editor’s Note, published May 1, said.
That’s a good practice. The newspaper’s “Principles for Using Generative AI in The Times’s Newsroom,” for the staff doesn’t contain express “do’s and don’ts,” unlike its policy for freelancers. Internal memos may provide more detailed advice for reporters and editors.
Neither statement requires that reporters or freelancers disclose the use of AI to editors.
When asked about disclosure, a spokesperson for the Times pointed to a paragraph in the principles for the newsroom which says in part: “We should tell readers how our work was created and, if we make substantial use of generative A.I., explain how we mitigate risks, such as bias or inaccuracy, with human oversight.”
Perhaps the Times’ thinking is: If we should tell our readers, then our reporters should tell their editors. But as these two episodes reveal, Times’ reporters are using AI and not telling their audiences or apparently their editors.
What’s needed are candid conversations about AI usage before publication among editors and reporters, communication managers and writers.
“Ongoing internal disclosure is important,” we wrote in 2024.
Disclose or not
When to disclose AI use is an unsettled question. Some AI users are reluctant to say they’ve done so at least in part because of the “rebellion” against the new technology.
A “Don’t ask, don’t tell” attitude is emerging.
This year, just 29% of in-house public relations professionals said they want their PR agency to disclose AI use all the time, down from 37% in 2025, according to a survey by media monitoring firm Muck Rack released in January.
Just 23% of PR agencies disclose AI use to clients, the survey of 564 PR people found. The Public Relations Society of America addressed disclosure for the first time when it updated its ethics guidance on AI in October 2025.
It now recommends: “Clearly disclose when content, decisions, or interactions are significantly influenced or generated by AI, especially when this information could impact how messages are perceived, how relationships are built, and how trust is maintained.”
Making that decision or the Times’ decision on what constitutes “substantial use of generative AI,” should be come only after candid conversations.
The damage to reputations can be too great.
Tom Corfman, a lawyer and senior consultant with Ragan Consulting Group, feels like he needs all the intelligence he can find. Does your organization need help establishing an AI policy? Email Tom to set up a free call with him and RCG co-founder and senior partner Jim Ylisela.
Follow RCG on LinkedIn and subscribe to our weekly newsletter here.
