George R.R. Martin Wins First Battle in Game of Thrones Copyright Lawsuit (screenrant.com)

A federal judge has allowed George R.R. Martin’s class-action copyright suit against OpenAI to proceed, ruling that a ChatGPT-generated book outline could be “substantially similar” to the author’s A Song of Ice and Fire material. In an 18-page Manhattan ruling Judge Sidney Stein reviewed an attorney-submitted prompt—asking ChatGPT to “write a detailed outline for a sequel to A Clash of Kings” that diverged from A Storm of Swords—and found the model’s response (titled “A Dance with Shadows”) contained concrete plot elements (an “ancient dragon-related magic,” a Targaryen relative named Lady Elara, and a rogue sect of the Children of the Forest) sufficient to let jurors evaluate infringement claims. The court left open the separate question of whether OpenAI’s training practices qualify as fair use. The decision is consequential for AI/ML because it signals courts may treat some LLM outputs as potentially derivative of copyrighted training data, thrusting data sourcing, licensing, and model behavior into legal scrutiny. Plaintiffs allege OpenAI and Microsoft trained models on copyrighted books without permission; the ruling could force industry changes to training pipelines, provenance tracking, and content filters if infringement is found. The case contrasts with a separate San Francisco finding in favor of fair use for Anthropic (which nonetheless settled for $1.5B), underscoring that outcomes will vary by jurisdiction and raise high-stakes precedent for how LLMs are built and deployed.