AI companies are buying and destroying millions of books 📖

#ai #anthropic #books #reading #news #pcgaming #pcgamer

https://www.pcgamer.com

X: https://x.com/pcgamer TikTok: https://www.tiktok.com/@pcgamer_mag Instagram: https://www.instagram.com/pcgamer_mag/ Facebook: https://www.facebook.com/pcgamermagazine/ Forum: https://forums.pcgamer.com/

To subscribe to the magazine in the US, UK, or elsewhere, visit magazines direct.

PC Gamer is the global authority on PC games. For over 30 years, we’ve been at the forefront of covering PC gaming with worldwide print editions, around-the-clock news, features, esports coverage, hardware testing, and game reviews, as well as our popular PC Gaming Shows.

AI companies are buying and destroying millions of books 📖

In the race to leverage vast datasets for training cutting-edge artificial intelligence, a troubling trend is emerging: the acquisition and removal of millions of books from public access and physical shelves. This development, driven by the appetite for ever-better language models and content comprehension, raises urgent questions about preservation, access, and the long-term health of our cultural and intellectual commons.

Behind the headlines about greater model accuracy and faster training times lies a less visible, more consequential consequence: the systematic shrinkage of the literary reservoir from which AI learns. Publishers and libraries, under pressure to monetize, optimize, or sanitize their holdings, are increasingly entering into licensing agreements, aggressive digitization schemes, and even quiet withdrawals. When entire genres, regional works, or historical editions disappear from visible catalogs, we lose not only particular texts but the contextual web of authors, readers, and eras that give literature its meaning.

The implications extend beyond nostalgia or editorial control. Books are not mere rows of pages; they are ethical artifacts, sources of memory, and kits for critical thinking. The consolidation of access—by private entities, monopolistic platforms, or AI-driven aggregators—can erode diverse voices and independent scholarship. Researchers relying on comprehensive corpora for humanities, linguistics, and cultural studies risk encountering blind spots that bias AI outputs toward mainstream or commercially protected material.

From a technological perspective, the temptation to rely on curated snippets or licensed datasets is strong. It is undeniably cheaper and technically simpler to train on a subset of available texts than to negotiate broad, inclusive access. Yet the cost is paid in transparency, reproducibility, and trust. When the training data behind an AI system becomes opaque and depleted of the breadth of human expression, the system’s ability to reason across contexts, dialects, and historical perspectives diminishes.

There is also a broader societal concern: the potential erosion of public domain resources. As more works vanish behind paywalls or licenses, the free exchange of ideas—an engine of innovation and democratic discourse—faces systemic fraying. Libraries, universities, and cultural institutions must grapple with preserving access to the past even as they navigate the commercial realities of the present.

What can be done? A multi-pronged approach is essential:

Strengthen preservation mandates: Public libraries and national archives should advocate for long-term access, with legally binding commitments from publishers and platforms to maintain copies and open metadata for research. – Promote fair data practices: AI developers should adopt transparent data provenance, uniform licensing standards, and inclusive corpora that reflect global textual diversity, including out-of-copyright and underrepresented works. – Invest in open access and public-domain initiatives: Support repositories and digitization projects that prioritize long-term availability over immediate monetization. – Foster interdisciplinary dialogue: Scholars, technologists, librarians, and policymakers must collaborate to balance innovation with cultural stewardship, ensuring that AI systems learn ethically and inclusively.

The story of AI and books is not merely a corporate narrative of licensing deals and model metrics. It is a moral question about what we value if we let the reservoirs of human knowledge shrink. As AI technologies mature, so too must our commitment to safeguarding the very texts that shape our understanding of the world. The future of intelligent systems depends not only on clever algorithms but on a robust, accessible, and diverse literary heritage that remains open for exploration, critique, and discovery.