The Battle for Books: AI's Destructive Scanning Dilemma
The recent court case of Bartz v Anthropic PBC has brought to light a startling practice in the world of AI language training: the destruction of books for the sake of data collection. Anthropic, an AI company, embarked on 'Project Panama,' a mission to destructively scan all the books in the world, as revealed in internal memos.
The rationale behind this controversial approach is intriguing. Anthropic needed a vast and high-quality language dataset to train its language model, Claude. Books, with their well-curated content and captivating narratives, seemed like the perfect solution. But instead of obtaining legal permission to use existing e-books, Anthropic opted for a more clandestine path.
They chose to use pirated sources, which eventually led to a hefty $1.5 billion settlement with authors. This decision raises questions about the company's ethics and respect for intellectual property. It's astonishing that a company would rather pay billions in settlements than navigate the 'legal/practice/business slog' of copyright management.
The Legal Loophole
The court's decision in this case is equally fascinating. It ruled that using proprietary material to train an LLM does not infringe on copyright, likening it to teaching a human to read and write. This interpretation sets a dangerous precedent, suggesting that AI companies can exploit creative works without consequence. It's as if the court is saying, 'If you can't beat them, join them' when it comes to AI's impact on intellectual property.
The Cultural Impact
Beyond the legal implications, this case highlights a profound cultural issue. Books have long been revered as cultural objects, carrying meanings beyond their textual content. Think of Donald Trump holding a Bible during the 2020 protests—a powerful symbol, regardless of the text's content. Yet, the US lacks legal provisions to protect its cultural heritage, especially books, from such destructive practices.
Preserving Human Authorship
The case prompts us to reconsider the value of human authorship. With AI-generated text becoming increasingly prevalent, should we start treating 'wholly human author' works as a category of cultural heritage in need of preservation? The fact that Anthropic is willing to destroy physical books rather than engage with their creators underscores the urgency of this question.
A Call to Action
Bartz v Anthropic PBC serves as a wake-up call. We must not devalue our roles as readers and writers, ceding control of our intellectual creations to AI companies. This case is a stark reminder that the battle for books is not just about physical copies but also about the power dynamics between human creativity and AI innovation. It's time to reclaim our cultural spaces and ensure that the means of production for our language lives remain in human hands.
