Vox Media and The Atlantic have signed licensing agreements with OpenAI, giving the AI company access to their content for model training and for display inside ChatGPT.
The agreements add two prominent publishers to a growing list of media organizations that have chosen to work with OpenAI. The terms were not disclosed, but the deals point to a larger shift: news content is becoming part of the infrastructure behind AI tools, not only material that readers find on publisher websites.
What the OpenAI agreements allow
Under the deals, OpenAI can use content from Vox Media and The Atlantic to train its AI models. Their material can also appear in ChatGPT, where users increasingly expect direct answers rather than a list of links.
When ChatGPT cites content from these publishers, it will include a source link. For the publishers, that may create a new path to visibility inside a chatbot interface. For OpenAI, it helps connect AI-generated responses to recognized media sources.
Vox Media will also use OpenAI's technology for its affiliate commerce and advertising data platforms. That part of the agreement shows that the relationship is not limited to content licensing. It also gives Vox Media a way to apply OpenAI tools inside business systems connected to advertising and commerce.
The financial terms of the Vox Media and The Atlantic agreements were not disclosed. OpenAI has previously signed deals with News Corp, Axel Springer, DotDash Meredith, the Financial Times and the Associated Press.
Why publisher content matters to ChatGPT
The source article presents two reasons OpenAI needs data deals. First, OpenAI needs content to train its AI. Second, it needs fresh information to help keep ChatGPT current, because ChatGPT is static on its own.
That second point is especially important for news. A chatbot that cannot reflect current reporting has limits as an information product. Licensing agreements can help OpenAI connect its systems to publisher material while also giving those publishers a formal place inside the product.
The article also notes that "SearchGPT, a Google Search competitor, seems very likely and could be launched soon." If that happens, the importance of publisher relationships could grow. A search-like OpenAI product would depend not only on answering questions, but also on deciding which sources are visible, linked and trusted.
OpenAI is also bringing its technology into newsrooms through the Newsroom AI Catalyst accelerator program, launched in partnership with WAN-IFRA. According to the source, 128 international newsrooms are participating.
The money question remains partly hidden
The terms of the Vox Media and The Atlantic agreements were not disclosed, so it is not possible to compare them directly with other OpenAI media deals from the information provided.
The source article says the News Corp deal is rumored to be worth $250 million. It also says the Springer deal is said to be worth tens of millions per year. Those figures suggest why publishers may be willing to negotiate with OpenAI, especially as AI platforms become more central to how people find information.
Still, the lack of disclosed terms matters. Without public details, it is hard to know what publishers are receiving, what rights they are granting, and how much control they retain over the use and presentation of their work.
Why the deals are controversial
The source article argues that OpenAI's media deals are worrisome because the company could become a major new internet platform. In that role, it may gain early control over media visibility while also reducing the number of potential legal opponents.
Many publishers are suing OpenAI for using their data to train its models without permission. Licensing agreements with major media companies could change that conflict by turning some publishers into partners, while others continue to challenge OpenAI in court.
The article includes a warning from Jessica Lessin of The Atlantic, described as a publisher who has now also decided to take OpenAI's money. She wrote: "Publishers should be patient and refrain from licensing away their content for relative pennies. They should protect the value of their work, and their archives. They should have the integrity to say no. It’s simply too early to get into bed with the companies that trained their models on professional content without permission and have no compelling case for how they will help build the news business."
That quote captures the central tension. Licensing may create revenue and visibility, but it may also weaken the collective bargaining position of publishers who believe their archives and reporting have been used without permission.
What happens when news moves into ChatGPT
Putting news content inside ChatGPT changes more than the route readers take to find information. It also changes the context in which that information appears.
On a publisher's own website, design, layout and user experience help signal identity. A tabloid newspaper may look bright and colorful, while a business newspaper may present itself differently. Inside a chatbot, those visual and editorial cues can disappear.
The source article argues that a report from one kind of publication could look just as legitimate as another when both appear inside ChatGPT. That raises a practical concern for media companies: if readers encounter their work mainly through AI interfaces, the publisher brand may become less distinct.
For Vox Media and The Atlantic, the OpenAI agreements create a formal role in this changing environment. For the broader media industry, they sharpen a larger question: whether licensing content to AI companies helps protect journalism's value, or accelerates a future in which publishers become suppliers for platforms they do not control.