Game of Thrones author George R. R. Martin is a well-known AI skeptic — he once called it "the world's most expensive and energy-intensive plagiarism machine," after all. And now he's making official accusations under penalty of perjury with a new lawsuit against OpenAI that hinges on a novel legal theory. According to the Hollywood Reporter, Martin asserts that ChatGPT infringes on his copyrights every time it answers a user prompt about the Game of Thrones books, whether it generates a plot summary or assembles a hypothetical plot line for the long-awaited sixth book in the series.
Martin, along with a number of other plaintiffs, allege that OpenAI infringed on their copyrights in multiple ways — training their AI models on the copyrighted works, pirating their books from so-called "shadow libraries" that were not intended for training, and generating answers to user prompts that are "substantially similar" to the copyrighted works that were used for training data.
In other words, when a ChatGPT user requests a summary of A Song of Ice and Fire, the resulting answer violates Martin's copyright. When a user tells ChatGPT to come up with a plot from the long-awaited (and still unpublished) sixth book, that violates his copyright as well.
So far, the courts have been amenable to these arguments, and the lawsuit will continue moving forward — and potentially help shape copyright law in our future AI-saturated world.
