It was revealed that Microsoft signed a three-year artificial intelligence training agreement with HarperCollins. This development was confirmed by Bloomberg shortly after 404 Media announced that HarperCollins had made a deal with an artificial intelligence company. Microsoft will use certain content from the publishing house to train a yet-to-be-announced artificial intelligence model.
In a statement to 404 Media, HarperCollins stated that this deal covers only a select number of “non-fiction back catalogue” titles and that authors must sign off before being included in the program.
On the other hand, screenshots shared by writer Daniel Kibblesmith on social media offered clues about the details of the agreement. Kibblesmith announced that he was offered $2,500 for each book in exchange for a three-year artificial intelligence license. However, Kibblesmith said that he rejected this offer and said in a post: “According to the result aimed with this technology, I will not accept it unless I am offered money that will not require me to work again.”
What will Microsoft do with HarperCollins content?
While Microsoft did not make a statement about what it would do with the model it developed using HarperCollins content, Bloomberg’s sources claimed that these materials would not be used to produce books written by artificial intelligence. The company refused to comment on this matter.
News Corp, the parent company of HarperCollins, made a similar deal with OpenAI earlier this year. In this context, OpenAI’s models can use News Corp’s digital content such as The Wall Street Journal, New York Post and The Daily Telegraph for training purposes.
The role of this agreement between Microsoft and HarperCollins in the company’s artificial intelligence development strategies and its sectoral effects are wondered. Authors, however, continue to demand transparency and fairer royalties in such agreements.