Live Feeds
● TRACKER Updated 22d ago · 11 sources tracked

Is it legal to train AI models on copyrighted books? It’s complicated

The legality of training AI models on copyrighted books is complicated and varies by jurisdiction. US courts assess AI training on copyrighted works differently, while some consider 'private use' as a potential safe harbor. AI companies have quietly built powerful models by ingesting millions of books, often without permission or payment.

🎙️

Listen to Live Briefing

Real-time synthesized voice briefing · Live Feeds Desk

⏱ ~2 min
Speed:
RSS Source map (15)
⚡ Key Developments & Real-Time Context
Text size:
  • ✓ AI models powering ChatGPT, Gemini, Claude, and other chatbots are trained on seemingly infinite databases of books.
  • ✓ US courts assess AI training on copyrighted works differently.
  • ✓ The use of 'private use' as a potential safe harbor for AI training is being considered.
🛡️ Source Corroboration: 11 independent reporting domains (75% confidence) ⏱ Read time: ~2 min

What changed

Recent developments indicate a shift towards settlements and FTC scrutiny in AI training legal battles.

Live updates

  1. Legality of training AI on copyrighted books uncertain

    The legality of training AI models on copyrighted books is complicated and varies by jurisdiction. US courts assess AI training on copyrighted works differently, while some consider 'private use' as a potential safe harbor. AI companies have quietly built powerful models by ingesting millions of books, often without permission or payment.

    Why it matters

    This issue is at the center of one of the most consequential copyright fights in decades, with billions of dollars in potential damages at stake. The controversy involves authors discovering their works had been swept up in massive datasets like Books3 and LibGen, used to train models including Meta's Llama and Anthropic's Claude.

    What is confirmed

    • AI models powering ChatGPT, Gemini, Claude, and other chatbots are trained on seemingly infinite databases of books.
    • US courts assess AI training on copyrighted works differently.
    • The use of 'private use' as a potential safe harbor for AI training is being considered.

    Still unconfirmed

    • Unlicensed, unrestricted AI training could destroy the ecosystem for books.

    What to watch next

    • Settlements and FTC scrutiny in AI training legal battles
    • Decisions on fair dealing and fair use in various jurisdictions
    • Evidence of AI companies obtaining permission or paying for copyrighted books
    Sources used for this update (11)
    1. TechCrunch — Is it legal to train AI models on copyrighted books? It’s complicated
    2. EUobserver — [Interview] Does copyright protect your AI-generated content in Europe? Let’s find out
    3. JD Supra — Who Owns the Copyright in Work Generated by an LLM?
    4. Columbia Undergraduate Law Review — Roundtable #36: Scraping By: Generative AI and the Limits of Cross-Border Governance
    5. Bar and Bench — Has “private use” become an AI safe harbour?
    6. www.androguider.com — Is Training AI on Copyrighted Books Legal? Fair Use, Lawsuits and Authors Rights Explained
    7. Supreme Court Observer — Public Interest, copyright and fair dealing: ANI v OpenAI
    8. ua.news — US courts assess AI training on copyrighted works differently — TechCrunch
    9. www.whalesbook.com — AI Training Legal Battles Shift to Settlements and FTC Scrutiny
    10. www.businessghana.com — Is it legal to train AI models on copyrighted books? It’s complicated
    11. tech.yahoo.com — 'Unlicensed, unrestricted AI training could destroy the ecosystem for books' — quote of the day by the Authors Guild on the sourcing of training data
    confidence 75%
📊

Community Sentiment: How do you assess this situation?

Voice your perspective · Real-time aggregated sentiment from the Live Feeds community