News Corp. seeks AI licensing deals for its content to keep generative AI up-to-date
Key Points
- News Corp CEO Robert Thomson reveals that the company is in advanced talks to license its content to generative AI models that need constant input to stay relevant.
- This follows the recent agreement between OpenAI and the Associated Press (AP) to license some of its data. AP also experiments with implementing OpenAI technologies in journalism.
- Meta has recently raised concerns about high licensing costs potentially stifling AI development. Google and OpenAI defend the transformative use of copyrighted data in AI models as fair use.
News Corp. CEO Robert Thomson expects significant revenue from licensing its content to companies for their AI models.
The company is in advanced talks with digital companies to use its content for generative AI models, which require constant input to stay relevant, Thomson said on an earnings call. News Corp reported revenue of $2.5 billion, up 1% from last year, with a net income of $58 million.
Generative AI engines are only as sophisticated as their inputs and need constant replenishing to remain relevant. And we are proud to partner with responsible purveyors of AI products and their prescient leaders.
Robert Thomson, News Corp. CEO
How many license agreements can be made?
There are ongoing discussions between AI model providers and publishers, the most high-profile of which involves the New York Times and may end in a lawsuit against OpenAI. The latter has already announced that it has reached an agreement with AP to license some of its data. Conversely, the AP is experimenting with OpenAI technologies, implementing them in the daily work of journalists without using them to generate new articles from scratch.
When OpenAI announced this deal, it seemed like a drop in the bucket to me. More recently, Meta confirmed my skepticism in its response to the U.S. Copyright Office's inquiry into AI and copyright, saying that licensing costs are too high for the amount of data needed to train large AI models.
Meta compares generative AI to tools like the printing press or the computer, and warns that high licensing fees could stifle AI development. It's not like generative AI is cheap to begin with, not even counting licensing costs.
Google and OpenAI also defended the transformative use of copyrighted data in training AI models in their responses to the Copyright Office. They argue that such use doesn't violate copyright law because it's not storing copyrighted data, but learning patterns from it, and that makes it fair use.
AI News Without the Hype – Curated by Humans
As a THE DECODER subscriber, you get ad-free reading, our weekly AI newsletter, the exclusive "AI Radar" Frontier Report 6× per year, access to comments, and our complete archive.
Subscribe now