Digital media publisher Ziff Davis has filed a lawsuit against OpenAI in Delaware federal court.

The company is alleging that OpenAI used its copyrighted content without authorization to train the AI model that powers ChatGPT.

The complaint, submitted on April 24, 2025, claims that OpenAI exploited Ziff Davis’ publications.

Ziff Davis asserts that OpenAI prioritized rapid development at the expense of the rights of content owners.

OpenAI has stated that its AI models are trained on publicly available data. The company operates under the principles of fair use.

While OpenAI commented on its stance, Ziff Davis declined to discuss the ongoing litigation.

🔗 Source: Reuters

🧠 Food for thought

1️⃣ Recent court ruling undermines AI companies’ fair use defense

A significant legal precedent is emerging that challenges the foundation of how AI companies obtain their training data.

The U.S. District Court recently ruled against Ross Intelligence in their dispute with Thomson Reuters, explicitly rejecting the argument that using copyrighted materials for AI training constitutes fair use 1.

The court found that copying legal content for AI training wasn’t transformative and served essentially the same purpose as the original works, a ruling that directly contradicts OpenAI’s defense strategy in the Ziff Davis case 2.

This decision particularly emphasized market impact as a critical factor, noting that AI-generated outputs could negatively affect the original copyright holders’ business—precisely what Ziff Davis argues regarding its nearly 2 million annually published articles 3.

The ruling marks a potentially dramatic shift in the legal landscape surrounding AI training data, challenging years of technology companies’ assumptions about permissible data usage.

2️⃣ AI training data shows strong preference for high-authority publishers

Research reveals AI companies disproportionately train their models on content from established media brands with high domain authority.

Recent OpenAI developments