AsianFin -- Artificial intelligence developers scored narrow legal victories this week as federal judges in California ruled that Anthropic and Meta could legally train large language models (LLMs) using copyrighted books.

The decisions mark a potential turning point for AI companies seeking legal clarity on the use of protected works in model training. However, the broader legal war over generative AI and copyright is just beginning.

Dozens of copyright holders have filed lawsuits arguing that developers must secure licenses before feeding creative works into AI systems for commercial purposes. They also contend that the AI-generated output must not resemble the original content too closely.

Rob Rosenberg, an intellectual property attorney at Telluride Legal Strategies, described the ruling in favor of Anthropic as “ground-breaking,” but cautioned that it represents “just the opening salvo in a much longer fight.”