Federal Judge Sidney Stein, of the U.S. District Court for the Southern District of New York (Manhattan), has ruled that the class action filed by author George R.R. Martin -creator of the A Song of Ice and Fire saga- together with authors Michael Chabon, Ta-Nehisi Coates, Jia Tolentino, and Sarah Silverman, against OpenAl and its partner Microsoft, may proceed. The Court found that the texts generated by the artificial intelligence system ChatGPT could be substantially similar to the plaintiffs’ original works and, therefore, might infringe the exclusive economic rights protected under U.S. copyright law.

The lawsuit is based on the alleged unauthorized use of protected literary works to train the defendant company’s language models. Among the evidence presented, the writers’ attorneys asked ChatGPT to draft an alternative sequel to A Clash of Kings, distinct from A Storm of Swords. The resulting text -titled A with Shadows -included Dance characters, plots, and narrative devices analogous to those of the original work, which the judge deemed sufficient for a reasonable jury to find substantial similarity. On that basis, the proceeding was admitted as a class action.

While the ruling does not yet address the merits of the dispute, it establishes that OpenAl’s defense, based on the doctrine of “fair use,” will be examined at a later stage of the proceedings. This principle allows for limited use of protected works for purposes such as research or technological development; however, its application in the context of artificial intelligence remains controversial. The mass reproduction of copyrighted material for algorithmic training raises complex questions about the boundaries of lawful use when the outputs generated replicate the creative expression of the original works.

The decision is particularly significant in recognizing that the training of artificial intelligence systems using copyrighted works may constitute a form of unauthorized reproduction. Should this reasoning be upheld in subsequent stages, it could substantially alter the way technology companies manage the materials used in the development of their language models. Such could lead to the establishment of licensing systems, prior authorization requirements, compensation mechanisms copyright holders.

Martin’s case joins other ongoing disputes concerning the use protected content in the technological domain. Recent claims-such as those filed by Disney and Universal against the image generator Midjourney, or the settlement reached by Anthropic to compensate affected underscore the need regulatory boundaries. Judge Stein’s decision, by allowing the case to proceed, reaffirms the growing judicial scrutiny over the relationship between artificial intelligence, human creativity, and the effective protection intellectual property rights.