The Authors Guild has warned that unlicensed AI training on books could destroy the publishing ecosystem, responding to reports that AI companies are scanning printed literature on an industrial scale to build large language models.
Large language models need vast amounts of data to reach basic competency, and AI companies have pulled from public web content, synthetic data, and printed books. Recent reporting revealed that some firms allegedly used pirate ebook sites to source training material, prompting writers, authors, and publishers to speak out.
"Unlicensed, unrestricted AI training could destroy the ecosystem for books in the long run, and copyright exceptions do not extend to undermining the very purpose of copyright law," the Authors Guild said in its licensing guidelines.
Industrial-scale book scanning
The Authors Guild's guidelines cover the legal questions around using published works in AI training and warn that continued data harvesting risks the existing book ecosystem. The organization has been vocal about the need for clear licensing agreements between publishers and AI companies.
The concern has sharpened in recent months as court documents revealed the scale of book acquisition by major AI labs. Anthropic ran a project codenamed "Project Panama," described in filings as an effort to "destructively scan all the books in the world." The company used the codename because "we don't want it to be known that we are working on this," according to the documents.
Anthropic needed a large, high-quality dataset to train its Claude model. Books offered relatively high-quality writing compared with typical online content, so the company set out to buy them in bulk - purchasing, scanning, and destroying physical copies in the process.
Why this matters for writers
Writers face a direct financial threat from unlicensed training. If AI companies can harvest books without payment or permission, they capture the value of an author's work without compensating the creator. That undermines the economic model that lets professional writers earn a living from their craft.
The Authors Guild's position offers a practical reference point: books are protected by copyright, and using them to train AI without a license is not a gray area. Writers should understand their rights and watch for licensing opportunities that pay for the use of their work - and push back when companies bypass those agreements entirely.
Your membership also unlocks:
