OpenAI plans to invest US$1-5 million per year to obtain news licenses

Two media company executives who have recently been in talks with OpenAI said that OpenAI is willing to invest between $1 million and $5 million per year to reach agreements with news publishing companies to obtain the rights to use news articles to train large language models.

The company is currently in talks with about a dozen media companies. But The Information believes that this amount is a bit insufficient even for small publishers and may make it difficult for OpenAI to reach an agreement.

In December, OpenAI said it had partnered with publishing giant Axel Springer, its first such partnership with a publisher. Axel Springer is a German multinational mass media company that owns brands such as Business Insider and Politico. Financial terms of the deal were not disclosed, but the executives said it was in the tens of millions of dollars.

Meanwhile, Apple, which is trying to catch up with OpenAI and Google in the field of generative AI, is also trying to strike content-use deals with publishers, an executive revealed. Apple is providing more funding, but also wants rights to use the content more broadly than OpenAI; they want to be able to use the content in future AI products in whatever way the company deems necessary.

Currently, Apple has signed approximately $50 million worth of contracts with publishers such as Condé Nast (which owns Vogue and The New Yorker) as well as NBC News and IAC (which owns The Daily Beast and Better Homes and Gardens). dollar agreement.

The New York Times, CNN, Reuters and The Verge's parent company Vox Media have all recently banned OpenAI's GPT crawler from accessing data. The New York Times also sued OpenAI and Microsoft Corporation in December, saying the two companies illegally used their copyrighted materials to train AI models. In addition, several well-known authors have also launched lawsuits against these companies.

Guess you like

Origin www.oschina.net/news/274326/openai-offers-publishers-little-1m
Recommended