Meta and Anthropic win legal battles over AI training with copyright law challenges ahead.
From Yahoo Finance: 2025-06-27 05:00:00
Federal judges in California ruled in favor of AI developers Anthropic and Meta, allowing them to “train” large language models on copyrighted books, but the battle over AI developers’ use of protected works continues. Dozens of copyright holders have sued, arguing for compensation and that AI output cannot resemble their original works. The rulings are groundbreaking but mark just the beginning of legal challenges as judges apply copyright law to AI systems. The rulings address questions on how far large language models can rely on protected works and may lead to appeals to the US Supreme Court. US copyright law grants creators exclusive rights to their works, and without a license, the use of copyrighted material by large language models is seen as infringement. However, fair use doctrine allows for certain exceptions, such as using material for commentary, critique, or transforming it into something new. Both Anthropic and Meta argued that training their models on copyrighted material was transformative and qualified as fair use. The outcome of these cases remains uncertain, with defendants needing to show that their use of copyrighted material did not disrupt the market for the original works. Anthropic faces additional legal challenges, including a suit from Reddit for scraping users’ personal data and using it to train their AI. The company also faces allegations of copyright infringement from music publishers for training their model on lyrics from popular songs. Willful copyright infringement can result in substantial fines, with potential statutory penalties for Anthropic exceeding $1 trillion. The company must also face claims from authors regarding their use of a pirated library of over 7 million books. The judge’s decision on class certification is pending, indicating that Anthropic is not exempt from legal consequences.
Read more at Yahoo Finance: Meta, Anthropic win legal battles over AI ‘training.’ The copyright war is far from over.