How the fair use clause is being applied to generative AI

Context & Relevance

• Access to diverse and voluminous training data (books, articles, web content) is central to improving Large Language Models (LLMs).
• This includes both public domain and copyrighted works—raising significant legal and ethical issues when used without permission.

Relevance : GS 3(IPR , Technology)

Central Legal Issue

• Key Question: Does using copyrighted material for LLM training—without authorisation—constitute copyright infringement?
• In the U.S., this hinges on whether the use qualifies as “fair use” under Section 107 of the Copyright Act.

Fair Use Doctrine – Four Factors

Courts evaluate fair use claims based on:

1. Purpose & Character: Is the use transformative (e.g., generating new knowledge vs reproducing existing works)?
2. Nature of Work: Factual works are more likely to be fair use than fictional/creative ones.
3. Amount & Substantiality: How much of the original was used?
4. Market Effect: Does the use harm the original’s market or potential licensing revenue?

Case 1: Anthropic PBC (Claude LLM)

• Used copyrighted books—some legally purchased, some from questionable sources—to train its GenAI.
• Court ruling:
  • Training with legally purchased books = Fair Use  (due to transformative use).
  • Copying from illegal sources = Not fair use ; court refused to grant blanket protection.
• Key takeaway: Court distinguishes between transformative use and how the data was acquired.

Case 2: Meta (LLaMA LLM)

• Sued by 13 authors for using illegally sourced books for training.
• Court ruling:
  • Training = Fair Use  (highly transformative).
  • Plaintiffs failed to prove market harm with empirical data.
  • Court did not penalise unauthorised downloading as a separate infringement (unlike Anthropic case).
• Judge acknowledged “market dilution” concern but said proof of harm was lacking.

Comparison: Anthropic vs Meta

Precedent Case: Thomson Reuters v. Ross Intelligence

• Court held no fair use because AI simply retrieved legal opinions (not transformative).
• Also competed directly with plaintiff’s product—thus hurting the market.

Emerging Legal Standards

• Courts seem to support transformative use in GenAI training—tilting toward fair use.
• But evidence of market harm will be crucial in future cases.
• Use of illegally sourced data may be treated as a separate violation—creating liability even if training is transformative.

Challenges for Plaintiffs

• Hard to prove “market substitution” or “licensing market harm.”
• LLM outputs are often not reproductions, but generated content—making infringement indirect and difficult to establish.

Implications Going Forward

• Unsettled legal landscape: Outcomes will vary case-by-case, based on data sourcing, model purpose, and market effects.
• Need for clearer copyright licensing frameworks and/or legislative clarity.
• Future rulings may hinge on empirical studies, including AI impact on creative economies.