By Shirin Ghaffary, Graham Starr and Brody Ford
OpenAI is in talks with CNN, Fox Corp. and Time to license their work, according to people familiar with the matter, in a growing effort to secure access to news content to build out its artificial intelligence products while facing allegations it’s ripping off copyrighted materials.
The startup behind ChatGPT, a tool that lets users quickly crank out text, code and other content with simple prompts, is seeking to cut deals with numerous producers of news, video and other digital media that can be used to make the AI chatbot more accurate, relevant and up to date. OpenAI is also battling lawsuits alleging copyright infringement.
OpenAI is discussing licensing articles from Warner Bros. Discovery Inc.’s CNN that it can use to train ChatGPT and also feature CNN’s content in OpenAI’s products, according to one of the people, who spoke on condition of anonymity to discuss private matters. CNN and Fox are negotiating not just around licensing text, but also video and image content, the people said. CNN and Fox declined to comment.
Time’s Chief Executive Officer Jessica Sibley said in a statement that the publisher “is in discussions with OpenAI and we are optimistic about reaching an agreement that reflects the fair value of our content.” OpenAI’s talks with the three publishers have not previously been reported.
OpenAI told Bloomberg News last week that it’s talking to dozens of publishers about licensing deals, but did not cite specific companies. These partnerships are key to OpenAI’s future as it’s balancing the need for updated, accurate data to develop its models with public scrutiny about where that data is sourced from. One of the companies the AI startup had been in talks with, the New York Times, sued OpenAI and Microsoft late last month for using the publication’s articles without permission.
In response to a request for comment on publisher talks, a spokesperson for OpenAI pointed to the company’s recent blog post, which referenced “continued collaboration with news organizations.”
“Our goals are to support a healthy news ecosystem, be a good partner, and create mutually beneficial opportunities,” OpenAI said in the blog post Monday, pushing back at the Times’ lawsuit. The company said it has “pursued partnerships with news organizations” to train its AI systems on “non-publicly available content” and show “real-time content with attribution” in ChatGPT.
OpenAI said it’s in discussions with the News/Media Alliance, a trade group which represents over 2,200 media outlets worldwide, “to explore opportunities, discuss their concerns, and provide solutions.” The AI startup has also been in conversations with Gannett, News Corp. and IAC, according to recent reporting from the New York Times.
Some other large media companies are prepared to enter talks with OpenAI. “We have had prior dialogue with a wide range of developers, including OpenAI, which we expect may now transition into commercial discussions about the use of our journalism to build and power their products,” Guardian News & Media, which publishes The Guardian, said in a statement.
OpenAI recently inked a multiyear licensing deal with Politico’s parent company Axel Springer SE for tens of millions of dollars, Bloomberg previously reported. In July, OpenAI announced an agreement with the Associated Press for an undisclosed amount.
Not all big publishers are rushing into negotiations with OpenAI, however. The Washington Post has not been in talks with OpenAI in recent months, according to a spokesperson for the publisher. One media executive, who spoke on condition of anonymity, said their company is considering taking legal action against OpenAI, similar to the New York Times.
A key concern for publishers is compensation. The Information previously reported that OpenAI has offered publishers $1 million to $5 million a year to license their articles. That range is considered too low for certain top publishers, according to people familiar with the matter. Some media companies are open to a range closer to what Axel Springer received, one of the people said.
One media executive, who asked not to be named to discuss private matters was skeptical that a productive agreement could be reached with AI companies until the courts clarify how copyright law applies to generative AI. Some in the industry are also calling on the US Congress to step in.
On Wednesday, a Senate Judiciary subcommittee held a hearing about the oversight of AI in journalism. During the hearing, Condé Nast CEO Roger Lynch urged Congress to issue rules saying that copyrighted content must require a license to be used for commercial generative AI. “Current Gen AI tools have been built with stolen goods,” he said.
Several people familiar with media negotiations also stressed the importance of how OpenAI will feature publisher content, and how much traffic would be referred back to media sites to increase their audience. OpenAI has said that one of the goals of its negotiations is to display real-time content from publishers with attribution.
The stakes are high for OpenAI to maintain access to copyrighted works. In a submission this month to the UK’s House of Lords, OpenAI said “it would be impossible to train today’s leading AI models without using copyrighted materials,” given how much online content is protected by copyright.
“Limiting training data to public domain books and drawings created more than a century ago might yield an interesting experiment,” the company said, “but would not provide AI systems that meet the needs of today’s citizens.”
