In an early February ruling, a US court concludes that the doctrine of « fair use » does not apply in the case between Thomson Reuters and generative AI developer Ross, and finds that training its product with an annotated compilation of court rulings constitutes copyright infringement. In our previous episode, we presented Meta’s arguments in the case launched by Kadrey and other writers, who believe that LLaMa’s training with their works constitutes an infringement of their copyright, and are seeking compensation. The case is still in its early stages and far from being decided, but the parties’ arguments have been presented. Meta, like all AI developers being sued by copyright owners in the no fewer than 20 lawsuits currently pending in the United States, argues that the use of works to drive LLaMa is authorized by the « fair use » exception, which in certain circumstances places limitations on the author’s exclusive rights over his or her work. This argument has just suffered a blow in another case, pitting Thomson Reuters against Ross: a February 11 decision by the District Court of Delaware has just denied application of the « fair use » exception. History Thomson Reuters v Ross will go down in history as the very first case to pit a rights holder – Thomson Reuters…
