ChatGPT developer OpenAI has offered newspapers and other publishers between US$1 million and US$5 million per year for licences to gain access to their articles to train its large language artificial intelligence (AI) models.
This is according to a report on The Information website, citing two executives who have recently been in negotiations with the AI company.
It was suggested this was too small even for modestly sized media owners and could make it difficult for OpenAI to add more deals.
After months of negotiation failed, the New York Times last month sued OpenAI and major backer Microsoft over alleged copyright infringement related to the unauthorized use of millions of newspaper articles to train AI models, including ChatGPT and Microsoft’s Copilot.
In negotiations, the NYT had been demanding fair compensation but said that no resolution has been reached.
The legal complaint seeks unspecified monetary damages, a permanent injunction to halt further infringement, and the destruction of AI models or training sets that incorporate the newspaper's journalism.
Along with the NYT, several other publishers including the BBC, CNN and Reuters have also blocked the likes of OpenAI from unauthorised crawling and scraping of their online content as part of AI development in recent months.
However, OpenAI has bagged some deals with major media outlets, such as Axel Springer, owner of Politico, Business Insider, Bild and Welt, which in December agreed to provide ChatGPT with content that it can use to train its large language models and provide real-time news summaries.
The Associated Press is also allowing OpenAI to train its models on its news stories.
The financial terms of these deals have not previously been revealed.