AI Copyright: Training-Data Lawsuits and Licences — 2026-09-20
Microsoft’s CEO has reportedly admitted in sealed court documents that AI training requires licensing, contradicting previous public stances, while German courts issued new rulings against Suno AI for using YouTube music data. Meanwhile, publishers are intensifying pressure on OpenAI to pay for content as the New York Times lawsuit approaches a critical fair-use ruling.
AI Copyright: Training-Data Lawsuits and Licences — 2026-09-20
Top developments
Microsoft Admits Licensing Need in Sealed Documents
Recent unsealed documents in ongoing litigation reveal that Microsoft’s CEO acknowledged that AI training on copyrighted material constitutes a licensing requirement, a significant shift from public "fair use" defenses. The documents also highlight internal warnings about severe traffic losses for media sites, undermining claims that AI training is purely transformative without market harm. This admission could severely impact the defense strategies of other tech giants relying on similar fair-use arguments.
German Court Rules Against Suno for YouTube Training Data
The Regional Court of Munich I (LG München I) ruled that Suno AI’s training on copyrighted music from YouTube violates German copyright law. The court found that memorized music works and similar AI outputs constitute infringement, marking a significant precedent against using scraped platform content for generative audio models. While the judgment is not yet legally final, it signals a strict interpretation of EU copyright directives regarding text-and-data mining exceptions.
Publishers Demand OpenAI Pay for Training Data
As the New York Times v. OpenAI case moves closer to a definitive ruling, major publishers are aggressively arguing that OpenAI can easily afford content licensing fees. The plaintiffs contend that the sheer scale of OpenAI’s revenue makes the "fair use" defense untenable, especially given the documented market harm to journalism. This push coincides with the DOJ’s recent non-binding brief supporting fair use, creating a complex legal landscape where economic arguments clash with doctrinal ones.
Japanese Book Distributors Investigated for Anthropic Sales
The Japan Book Publishers Association (JBP) has sent a formal inquiry to Nippan, a major book distributor, after reports emerged that it sold large volumes of books to Anthropic for AI training. This development highlights the supply-chain risks for AI developers who acquire physical or digital books through intermediaries rather than direct licensing. It raises questions about whether such sales violate resale price maintenance agreements or constitute unauthorized reproduction for machine learning purposes under Japanese law.
Local view
Germany: Local tech media outlets like Borncity and Datenschutzticker are closely tracking the Munich court's decision against Suno, emphasizing the distinction between "memorization" in model weights and direct output reproduction. Legal experts cited in these reports argue that the ruling effectively closes the door on using scraped YouTube content without explicit licenses, potentially forcing AI music startups to pivot to licensed datasets or synthetic data.
Japan: Yomiuri Shimbun and Nikkei are covering the first Japanese lawsuit explicitly questioning whether AI-generated images can hold copyright, a foundational question for the country's creative industries. Additionally, the scrutiny of Nippan’s dealings with Anthropic has sparked debate among local lawyers about the legality of bulk book sales for AI training, with some arguing it bypasses individual author consent required under Japan’s Article 30-4 exception.
Context & numbers
US copyright lawsuits against AI companies have surged to approximately 140 cases by September 2026, a sharp increase from 45 cases recorded in June 2025. This exponential growth reflects a strategic shift by rights holders to flood courts with precedents before comprehensive legislation is enacted.
On the radar
- NYT v. OpenAI Ruling: A definitive judge's ruling on fair use in the Manhattan court is anticipated soon, following the exchange of dueling briefs and the DOJ's intervention. This decision will likely set the tone for all pending US AI copyright cases.
- Anthropic Settlement Details: Further disclosures regarding the Bartz v. Anthropic $1.5 billion settlement are expected to clarify how per-work compensation is calculated, providing a benchmark for future licensing deals.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Powered by
