Implications of Generative AI on Copyright Laws

Three significant U.S. court rulings in 2025 — Thomson Reuters vs Ross Intelligence, Bartz vs Anthropic, and Kadrey vs Meta — have offered legal clarity on the applicability of copyright laws to generative AI models and their training datasets.

1. Understanding Generative AI and Copyright Concerns

• Nature of Generative AI
  • Generative AI models are trained on vast datasets, including copyrighted and public domain content.
  • They can generate text, images, and other media that may mimic or closely resemble original works.
• Copyright Implications
  • Potential for reproduction of original works.
  • Raises questions on ownership, fair use, and infringement.
  • Concerns about economic losses for copyright holders if AI substitutes original content.

2. Legal Principles Involved

• Fair Use Doctrine (U.S. Jurisdiction)
  • S. courts have ruled that transformative use — where AI ‘learns’ from copyrighted texts — can fall under ‘fair use’.
  • However, training on pirated content is not shielded under this provision.
• Text and Data Mining (EU/UK Context)
  • Exceptions provided under specific circumstances, especially for non-commercial research or archival purposes.
• India’s Legal Framework
  • Governed by Copyright Act, 1957, particularly Section 14 (economic rights) and Section 52 (fair dealing exceptions).
  • India recognises copyrights for legal persons and supports international treaties (e.g., Berne Convention, TRIPS).

3. Key U.S. Court Rulings: 2025

• Anthropic Case
  • Judge William Alsup ruled that AI training using copyrighted works was transformative and comparable to human learning.
  • However, training using pirated data must face trial — piracy remains outside protection.
• Meta Case
  • Judge Vince Chhabria held that Meta’s training did not harm the market potential of original works.
  • Encouraged equitable compensation mechanisms but upheld ‘fair use’ defence.

4. Indian Context and Emerging Litigation

• ANI vs OpenAI
  • Expected to clarify how India’s Copyright Act is interpreted in AI contexts.
  • The central issue: whether generative AI's output violates exclusive economic rights of original copyright holders.
• Policy Ambiguities
  • No AI-specific provisions exist under current Indian IPR law.
  • Concerns about digital circumvention, piracy, and authorship standards remain unresolved.

Challenges in Legal Harmonisation

• Global Legal Divergence: No unified international consensus on AI and copyright laws.
• AI Authorship Debate: Traditional copyright requires human authorship, leading to legal uncertainty on machine-generated works.
• Pirated Content and Market Harm: Ongoing concerns about large-scale unauthorised data scraping.