The nation’s most influential news organization has spent the summer agonizing about artificial intelligence. “Do not put any proprietary information, including published or unpublished Times articles, into generative AI tools, such as ChatGPT, Bing Chat, Bard or others,” New York Times deputy managing editors Sam Dolnick and Steve Duenes, and director of photography Meaghan Looram, wrote in an email to newsroom and opinion staff on June 27. That includes “notes from your reporting…internal financial or audience data, or code from our products or stories” the management team said. “Do not use generative AI tools in any aspect of our journalism without getting approval, until we further explore the opportunities and the risks they bring,” continued the memo, which noted that “the public terms of use for almost all of these tools also carry significant legal risks for protecting our intellectual property and other rights.” Even before the email went out, I’m told, management had started clamping down internally, warning some desk heads directly about putting any articles or reporting into AI models. “They’re freaking out,” said one Times staffer.
The Times is not alone: Top executives like News Corp CEO Robert Thomson and IAC’s Barry Diller have been publicly sounding the alarm over AI for months. At this point most media managers have likely sent memos to staff about the developing technology; newsroom unions are contemplating the labor implications; legal and business departments know their IP has probably been used to train the models without compensation. “That’s already done. They’ve already robbed the candy store,” as another Times staffer put it. Still, the broader question—and fear—remains: What will generative AI do to the professional news industry? “I think, correctly, the Times is deathly afraid of what this can mean,” the staffer said. “It’s potentially an existential moment for the Times and for the news industry, so I think leadership is properly taking a very robust look at this. But what are we doing? No one knows.” (The Times declined to comment.)
The Gray Lady’s effort to address AI started back in the spring, with chief product officer Alex Hardiman leading the corporate effort while Dolnick and other senior editors lead the editorial front, according to a source familiar. The Times has dedicated roughly 60 staff in the newsroom to address the threats, and possible benefits, of AI in news. These staffers (most of whom are participating in these AI working groups on top of their regular jobs) are brainstorming, among other things, areas the technology could be used for in the newsroom, as well as ways to ensure the paper’s human-led reporting can remain distinct at the *Times—*particularly in a world where more news is written by AI, according to another source familiar. (The technology was also top of mind during the Times’s Maker Week, an annual event soliciting ideas from people across their workforce: using AI for chatbots in the Cooking section and for gift finders on Wirecutter were two ideas presented during the event last month.) The working groups are scheduled to convene on August 17 for a meeting, which I’m told has some 80 people on the invite list.
So far the Times has kept its AI deliberations internal. Semafor recently reported that the paper is not part of a coalition of media organizations hoping to negotiate with tech companies over how artificial intelligence uses their content, an effort IAC is leading. News Corp, as I previously reported, is also not part of the coalition, though Thomson recently said that the company is in active discussion with AI and tech companies “to establish a value for our unique content sets and IP that will play a crucial role in the future of AI.” The Associated Press cut its own deal with OpenAI—a two-year agreement to share access to certain news content and technology that marks one of the first official news-sharing agreements between a major US news company and an AI company. Meanwhile, the Times updated their terms of service with restrictions on data scraping.
The Times is proceeding cautiously. “Our approach of innovating strategically, rather than chasing the trend of the moment, has served us well and remains a blueprint for how we intentionally complement human expertise with digital tools,” Hardiman and Dolnick wrote in another internal memo this summer. “We’re keenly aware of previous moments of techno-euphoria that barreled past red flags that only later became obvious. In this case, the risks—for society, journalism and our own business—are starkly clear from square one, and we need to balance our enthusiasm with the sober reality that we need to work through the legal, journalistic and business implications of these tools before we can put them into practice.”