Vox Media President Pam Wasserstein sent a Slack message and an email to her staff on May 29 detailing what company journalists say is shocking news: Vox has signed a content licensing deal with OpenAI. were
The deal gives the AI company access to Vox's entire archive of existing content as well as its journalistic work to train ChatGPT and other models. Wasserstein sent the alerts moments before Axios published an exclusive detailing the licensing and product deal, surprising its reporters.
The Atlantic writers, who signed a similar deal with the Microsoft-backed AI giant, were also not given the nod.
A May 30 statement from Atlantic Union read, “Atlantic staff learned about the deal largely from outside sources, and both the company and OpenAI declined to respond to questions about the terms of the deal. has refused.”
None of the current or former journalists at any of the companies that TechCrunch interviewed had any idea that their work would be outsourced to OpenAI. All of them are concerned that their employers are making short-sighted deals that will ultimately hurt writers and journalism as a whole.
Both Vox Media — which includes publications like The Verge, New York, Eater, The Cut and more — and The Atlantic have published pieces critical of OpenAI and generative AI. Eater reporter and Vox Union communications chair Amy McCarthy said she expressed concerns about the environmental impact of the power required to run large language models, board changes to OpenAI, and a “general lack of trust” in the company. What is it. .
Vox did not respond to a request for comment.
Since the deals were announced, journalists from each publisher have met with business-side executives to learn more about the deals, looking for one key piece of information: What's in it for journalists?
A sense of urgency
In the face of an increasing number of AI media deals, news guilds are now accelerating the pace of negotiations to replace AI protections fought for by Hollywood writing teams.
“The Writers Guild and Vox Media Union strongly believe that the implementation of AI is an essential subject of bargaining, even though our contracts do not explicitly contain AI provisions,” McCarthy told TechCrunch. “There are provisions in our contract that basically mean that the company has to bargain with us on fundamental changes in our working conditions, and we very much believe that this is a workplace issue, that this work is the issue of the circumstances, and that the company is obliged to bargain with us as to how it will work.
This means that striking publishers with AI providers may be contractually required to engage in discussions and negotiations with unions about these changes.
Atlantic Media Union had also planned to bring the issue to the bargaining table, but the OpenAI deal added a sense of urgency, a current employee told TechCrunch, requesting anonymity.
During negotiations this month, the Atlantic union put forward a proposal that would not allow AI to be used for writing, fact-checking, copy-editing and copy-editing. He also suggested that authors can use AI at their discretion in accordance with journalistic principles and ethics, but cannot be forced to use it. The proposal is yet to be accepted.
Other unions are working for similar protections. Nebraska journalists at the Omaha World-Herald Guild received protections from AI earlier this year. In 2023, after CNET published a series of AI-powered articles, the publication's journalists went public with their union drive, demanding AI protections and saying that AI in employee workflows How is it implemented?
It is important for companies to include such protections in journalists' contracts, as protection by law is not guaranteed. Companies like OpenAI claim that they are not breaking copyright laws by removing publicly available content. They also say that their chatbots don't fully reproduce content.
But publications like The New York Times, Raw Story, AlterNet and The Intercept have sued OpenAI for using copyrighted works by journalists to train ChatGPT without legally crediting or citing the sources. . Novelists, computer programmers and other groups have also filed copyright lawsuits against OpenAI and other companies developing generative AI.
Richard Toffel, former president of the nonprofit newsroom ProPublica and a consultant to news outlets, thinks the cases will end up in the Supreme Court. If the courts decide that OpenAI and others are guilty of copyright infringement, “they will need to enter into agreements with everyone else.”
Tofel believes that most publishers will make deals with AI companies. He noted that Google faced similar copyright suits when its search product was launching, but by the time they were resolved, users were so dependent on search that any publisher would not be able to access its content. He didn't want to keep her away from him.
McCarthy says the authors can't just rely on the courts: “We have to look at every possible avenue as a way to push back against the implementation of AI.”
Another concern for journalists is the adoption of AI by publishers to write content, which some media outlets have already started experimenting with.
CNET and Gannett have published AI-generated stories and art, and in the case of Sports Illustrated, under fabricated bylines. These stories were called out as AI-generated primarily because they were riddled with factual errors, but if AI gets a free pass for training in good journalism, these glaring errors should diminish over time. can
Journalists will not ask this question, so who will?
Journalists understand the basic structure of the deals, but they still have questions.
The Atlantic's VP of communications, Anna Bruce, said the company's partnership positions it as a premium news source within OpenAI, similar to deals with other publishers.
“Articles from The Atlantic will be discoverable in OpenAI products, including ChatGPT, and as a partner, The Atlantic will help shape how news is presented,” Bross told TechCrunch. And how that's presented in real-time discovery products in the future,” Bross told TechCrunch. “This agreement ensures how our content appears in OpenAI's products. … If an Atlantic article appears in response to a question, there will be Atlantic branding and a link to the article back on our site. Will be.
Bruce notes that this is not a syndication license, meaning that OpenAI is not permitted to reproduce articles from The Atlantic or similar reproductions of entire articles or long excerpts.
However, Atlantic journalists are still waiting for their leadership to explain why such content does not qualify as derivative work, for which they would have the opportunity to be paid directly. The Atlantic recently launched a new line of paperback books with collected works by its authors, and it has compensated authors for derivatives, multiple sources told TechCrunch.
The Atlantic's editorial staff raised the topic at an all-party meeting chaired by the publication's CEO Nick Thompson in mid-June, and learned that when ChatGPT gained access to their work, The edit team is otherwise “fairly insecure. This.”
In other words, there is no immediate danger of using ChatGPT to write articles.
The financial terms of The Atlantic and Vox's contracts still don't leave journalists in and out of the publications, but we do know they are two-year deals and will include using OpenAI technology to build products and features. OpenAI says its tech won't be used to imitate authors' own voices.
News Corp., the Wall Street Journal's parent company, also signed a deal with OpenAI that is reportedly worth more than $250 million over five years. Axel Springer, who runs Politico and Business Insider, has also struck a deal with OpenAI worth tens of millions of euros.
Other media outlets that have already signed similar partnerships with OpenAI include Dotdash Meredith (publisher of People, Better Homes & Gardens, Allrecipes, Investopedia and more), The Associated Press, Financial Times, France These include Le Monde, and Pressa Media in Spain.
(We should also note that TechCrunch's parent company, Yahoo, is also working with AI through the Yahoo News app. The app is powered by the core code of Artifact, which Yahoo acquired in April.)
OpenAI claims its contracts will help journalists drive traffic back to their articles, but that remains to be seen as the implementation is not yet live.
If users could ask an AI chatbot for the latest information on the Israel-Hamas war, for example, it would present “the ultimate nightmare for news companies,” Toffel said.
“They can be eliminated very significantly by an AI news product,” he said.
OpenAI was not able to verify details about user experience design, which may determine how likely a reader is to click on an article's external link.
And if readers don't have to go to a publisher's website to read articles, its ad revenue will suffer — something the news industry already struggles with as Google and Meta expand their News is prioritized in the algorithm. Journalists and writers will also have a smaller audience for their work.
Journalism suffers from a lack of funding, largely because tech companies like Meta and Google are taking a large share of digital ad revenue today. Publishers will no doubt welcome new revenue streams to bolster their balance sheets.
But journalists are questioning whether this is the best way.
“It feels like a lot more of a protection racket,” McCarthy said. “Like we made a deal with the guy who just burgled our house, and he's making a pinky promise that he's not going to burgle.” ”
Some AI startups are already picking up content without a contract. For example, ChatGPT rival Perplexity is under fire from Forbes for plagiarism, and Wired recently found that the AI company was secretly scraping its website. Despite these claims, Perplexity is set to announce ad revenue-sharing deals with publishers next week, the startup told TechCrunch.
Still, it looks like we can expect more deals like this in the future as publishers all seem to come to the same conclusion: AI is going to steal our jobs anyway. It can be paid for as well.