NYT v. OpenAI

New York Times v. OpenAI & Microsoft — The Biggest AI Copyright Case

Case Summary

The New York Times sued OpenAI and Microsoft in December 2023, alleging that ChatGPT was trained on millions of NYT articles without permission, creating a product that directly competes with and substitutes for NYT journalism. This is widely considered the most significant AI copyright case currently active.

Timeline

Key Legal Issues

Is AI Training on News Articles Fair Use?

The central question: Can OpenAI legally copy millions of copyrighted articles to train a model that generates competing content?

NYT's Arguments:

• ChatGPT can reproduce NYT articles nearly verbatim
• The AI substitutes for NYT subscriptions (market harm)
• Training involved wholesale copying of copyrighted works
• OpenAI profits commercially from NYT's investment in journalism

OpenAI's Arguments:

• Training is transformative use (learning patterns, not copying)
• Similar to Google Books scanning (found to be fair use)
• AI outputs are original, not copies
• NYT benefits from being referenced by AI

Why This Case Matters

1. Scale: Millions of articles, billions in potential damages

2. Precedent: Could define whether AI training on copyrighted content is legal

3. Industry impact: Every AI company trains on web content

4. Media implications: Could reshape how news organizations interact with AI

The Verbatim Reproduction Issue

NYT demonstrated that ChatGPT can reproduce substantial portions of NYT articles word-for-word when prompted. This is significant because:

• It suggests the model memorized copyrighted content
• Verbatim reproduction is harder to defend as "transformative"
• It demonstrates direct market substitution

Potential Outcomes

If NYT Wins:

• AI companies may need to license training data
• Could trigger wave of similar lawsuits from publishers
• Training data costs could reshape the AI industry
• Precedent for all content creators

If OpenAI Wins:

• Validates AI training as fair use
• Content creators have limited recourse
• Accelerates AI development without licensing friction
• But may be limited to specific facts of this case

Related Developments

• Copyright Office Part 3 (May 2025): AI training not always fair use
• Judge expressed skepticism about blanket fair use defense
• Multiple other publishers have filed similar suits

Current Status

ACTIVE — In discovery phase as of April 2026. Trial date not yet set. Settlement discussions reportedly ongoing but no agreement reached.