During the year, several legal disputes have taken place between tech companies and people in creative industries who believe that the unauthorized use of their works to train AI models is a violation of copyright.
Last week, a federal American court ruled that Facebook owner Meta was entitled to use copyright-protected books to train its AI system, but at the same time stated that it had violated copyright by using pirated books.
According to Mikaela Zabrodsky, CEO of the Swedish Publishers' Association and a lawyer specializing in intellectual property law, a large part of the books that AI companies train on are pirated.
It is obvious that there seems to have been very little, if any, respect for the rights holders whose works have been used, she says to SvD.
The Swedish Publishers' Association is now discussing how to act if specific Swedish books are discovered to have been used for AI training.
We know that there are Swedish works in these datasets that consist of pirated books. We have not quite landed on what we should do about it. But taking legal action is something that is in the balance for both publishers and authors, says Zabrodsky to the magazine.
