News organizations must have a seat at the table
In a moment that may quietly reshape the relationship between human storytelling and machine intelligence, the Associated Press and OpenAI have entered into a landmark licensing agreement — the first of its kind between a major news organization and an AI company. AP offers decades of vetted, factual reporting as training material; OpenAI offers its technology and expertise in return. The arrangement reflects a broader reckoning unfolding across the news industry: whether to resist generative AI's encroachment or to help steer it toward something trustworthy.
- Newsrooms have watched generative AI with deep unease — its tendency to fabricate convincing falsehoods cuts directly against journalism's core obligation to truth.
- The question of whether readers can even tell human reporting from machine-generated text has created an identity crisis for an industry already under financial strain.
- AP is betting that by licensing its archive and gaining a seat at the table, it can help shape AI development rather than simply absorb its disruptions.
- OpenAI gains something it urgently needs: high-quality, fact-checked, institutionally vetted content to train systems that have been criticized for unreliability.
- The deal's financial terms remain undisclosed, and how AP will actually deploy OpenAI's tools in its newsroom is still undefined — leaving the partnership's real-world impact an open question.
- If it holds, this agreement could become the model other publishers follow, turning archival journalism into a new kind of currency in the AI economy.
On Thursday, the Associated Press and OpenAI announced the first major licensing deal between a prominent news organization and an artificial intelligence company. AP will provide OpenAI with access to a portion of its historical news archive to train AI systems including ChatGPT; in exchange, AP gains access to OpenAI's technology and product expertise. Financial terms were not disclosed.
The partnership arrives amid genuine unease in journalism. Generative AI has impressed the public with its versatility, but newsrooms have been slow to embrace it — troubled by the technology's well-documented tendency to produce plausible-sounding falsehoods, and by the deeper challenge of maintaining reader trust when the line between human and machine authorship blurs. Some outlets have experimented at the margins, but AP's move is different in scale and intent.
By licensing its archive, AP is asserting that its decades of edited, fact-checked reporting holds real value as training material — and that news organizations should help shape AI rather than simply be shaped by it. AP's chief revenue officer framed the deal as securing a seat at the table for newsrooms of all sizes. OpenAI's COO, for his part, acknowledged that building trustworthy AI requires training it on trustworthy information.
Whether this becomes a template for the broader industry or a cautious one-off experiment remains to be seen. For now, two industries long wary of each other are wagering that a path forward together is better than navigating the uncertainty alone.
On Thursday, the Associated Press and OpenAI announced a partnership that hands over a portion of the wire service's vast archive of news stories to the artificial intelligence company. The deal marks the first major licensing agreement between a major news organization and an AI firm, and it signals how the two industries—long wary of each other—are beginning to find common ground.
Under the arrangement, AP will gain access to OpenAI's technology and the company's product expertise. OpenAI, in turn, gets to use AP's historical news content to train its AI systems, including ChatGPT, the chatbot that has captivated millions of users since its public launch. The financial terms were not disclosed. Neither company revealed specifics about how AP would actually deploy OpenAI's tools in its newsroom operations, though the wire service has already been experimenting with automation for years—using AI to generate corporate earnings summaries, recap sports games, and transcribe certain live events.
The partnership arrives at a moment of genuine tension in the news industry. Generative AI systems have dazzled the public with their versatility: they can plan trips, summarize legal documents, write code, and compose essays. Yet newsrooms have moved cautiously. The technology has a documented problem with "hallucination"—generating plausible-sounding information that is simply false. There is also the harder question of how readers will know whether they are reading words written by a journalist or words assembled by a machine. Some outlets have begun experimenting anyway. BuzzFeed announced it would use AI to generate personality quizzes. The New York Times created a Valentine's Day message generator powered by ChatGPT. But these are narrow applications, not core journalism.
AP's move is different in scope. By licensing its archive, the wire service is essentially saying that its decades of reporting—stories vetted by editors, fact-checked, and published under the AP's reputation—has value as training material for AI systems. Kristin Heitmann, AP's senior vice president and chief revenue officer, framed the deal as a way for news organizations to shape the technology rather than be shaped by it. "Generative AI is a fast-moving space with tremendous implications for the news industry," she said. "News organizations must have a seat at the table so that newsrooms large and small can leverage this technology to benefit journalism."
Brad Lightcap, OpenAI's chief operating officer, echoed that sentiment from the other side. He said AP's feedback and its archive of high-quality, factual reporting would help improve OpenAI's systems. The implication is clear: if you want an AI that generates accurate information, you need to train it on accurate information. AP's decades of published work—stories that have been edited, fact-checked, and published under institutional scrutiny—represents exactly that kind of data.
What happens next remains unclear. The deal could become a template for other publishers to follow, establishing a new revenue stream for news organizations while giving AI companies access to the kind of reliable training data they need to build more trustworthy systems. Or it could prove to be a one-off experiment, a way for AP to hedge its bets while the industry figures out whether generative AI is a tool for journalism or a threat to it. For now, the two sides are betting they can find a path forward together.
Notable Quotes
Generative AI is a fast-moving space with tremendous implications for the news industry. News organizations must have a seat at the table so that newsrooms large and small can leverage this technology to benefit journalism.— Kristin Heitmann, AP's senior vice president and chief revenue officer
AP's feedback and access to their high-quality, factual text archive will help improve the capabilities and usefulness of OpenAI's systems.— Brad Lightcap, OpenAI's chief operating officer
The Hearth Conversation Another angle on the story
Why does AP's archive matter so much to OpenAI? They could scrape the internet for training data.
They could, and they have. But there's a difference between quantity and quality. AP's stories have been edited, fact-checked, published under institutional pressure. That's signal in a sea of noise.
So AP is essentially selling its credibility?
Not selling it—licensing it. And in exchange, AP gets access to tools it can use in its own newsroom. It's a trade, not a one-way extraction.
But doesn't this create a conflict? AP is helping train the very systems that might put journalists out of work.
That's the real tension nobody's resolved yet. AP is betting it can shape how the technology develops rather than be crushed by it. Whether that works depends on what happens next.
What's the biggest risk here?
That the AI gets good enough at generating plausible news that readers can't tell the difference. That's the nightmare scenario—not that AI replaces journalists, but that it floods the zone with convincing misinformation.
And the upside?
If AP can actually use these tools to do better journalism faster—fact-checking, research, summarization—then maybe the partnership makes sense. The question is whether that's what actually happens or just what both sides hope will happen.