Authors can sue OpenAI for copyright infringement and shadow libraries, judge rules -.

Ever since it was discovered that OpenAI’s ChatGPT models were trained from copyrighted material without the permission of copyright holders, those holders have sought to sue Sam Altman’s company. The case that seems to have progressed the furthest is one led by Game of Thrones author George R.R. Martin, who is now considering a three-pronged approach in court along with other authors.

As noted by The Hollywood Reporter, the various arguments argue that training AI models on copyrighted material constitutes infringement, that generated responses from ChatGPT are too similar to the material they were trained on, and that the material is illegally copied from shadow libraries.

U.S. District Judge Sidney Stein recently allowed the separation of shadow libraries from copyright law. “The earlier class complaints alleged a cause of action for copyright infringement and alleged that OpenAI had impermissibly downloaded and reproduced plaintiffs’ books.” wrote the judge. “The fact that many of the allegations in the earlier class complaints suggested that the ultimate goal of the reproduction was to train OpenAI’s LLMs is not dispositive.”

Plaintiffs now have a few avenues ahead of them to seek damages, but it seems this is about more than the money, as this case could set a precedent for how AI handles copyrighted material in the future.