The American daily New York Times filed a lawsuit on Wednesday in a federal court in New York against OpenAI, the inventor of the ChatGPT software, and Microsoft, its main investor, for copyright infringement. Author.
• Also read: “Threat to humanity”: OpenAI employees are reportedly concerned about the rapid development of new AI
“As detailed in the complaint, Microsoft and OpenAI used our work to develop and market their generative artificial intelligence (AI) products without the Times' permission,” a spokesperson for the daily told AFP. It is actually a “copyright violation”. content and journalistic work”.
According to the lawsuit, the newspaper estimates the actual and legal damage caused to be “several billion dollars.”
“Defendants’ generative AI relies on massive learning models created by copying and using millions of copyrighted Times articles,” the paper added.
Additionally, Microsoft and OpenAI have “placed particular emphasis on Times content in building their learning model, emphasizing a preference that recognizes the value of this work.”
The daily also explains that it has tried to negotiate with the two companies to “obtain appropriate compensation for the use of their content”, but so far without success.
“Copyright laws protect our journalism. By law, if Microsoft and OpenAI want to use our work commercially, they must first obtain permission. “They didn’t do it,” emphasized a spokesman.
Interviewed by AFP, neither Microsoft nor OpenAI have yet commented.
In mid-December, OpenAI agreed to a payment with the German group Axel Springer, specifically publisher of the tabloid Bild, to provide the group's content in response to requests from ChatGPT users.
As part of this partnership, presented by the two players as unprecedented, users who ask ChatGPT a question will receive in response summaries of articles published by Axel Springer's brands, in particular Politico, Business Insider and the daily newspapers Bild and Welt became.
Since the launch of ChatGPT a year ago, the giants of Silicon Valley have been in a hectic race for so-called generative AI. This makes it possible, upon simple query in everyday language, to obtain texts, images or lines of code equivalent to those of humans, models that are trained primarily thanks to the mass of content available online.