Source: The Information Difference

This article examines the complex legal landscape surrounding the use of copyrighted materials in training large language models (LLMs) for AI. It highlights the extensive data sources used, including publicly available datasets and unauthorized scraping of social media, books, and media content. The longstanding copyright laws dating back centuries are challenged by AI companies’ claims that training constitutes fair use.

Several high-profile lawsuits are underway, including Bartz v Anthropic in the US and a case against Google in the EU, with potential class-action status and significant financial risks for AI developers. The article also discusses the problem of AI-generated “hallucinations” in legal documents, where fabricated citations have led to sanctions against lawyers.

Governments worldwide are beginning to regulate AI, adding layers of complexity. The article underscores the need for legal vigilance and careful control of AI use, particularly in the legal profession, to mitigate risks.