AI Copyright: Training-Data Lawsuits and Licences — 2026-09-12
The legal landscape for AI training data intensified this week as the US Department of Justice formally backed OpenAI’s fair use defense, arguing that barring such training would severely hamper technological progress. Meanwhile, publishers including the Seattle Times and Newsday filed new lawsuits demanding the destruction of AI training models, while the Authors Guild pushed for a summary judgment in its class action against OpenAI. In Europe, German courts ruled against OpenAI in a separate GEMA song lyric case, highlighting the diverging global regulatory approaches.
AI Copyright: Training-Data Lawsuits and Licences — 2026-09-12
Top developments
DOJ Argues Barring AI Training Would Hinder Progress
In a significant policy intervention published around September 4, 2026, the US Department of Justice (DOJ) filed an amicus brief supporting OpenAI in the New York Times lawsuit. The DOJ argued that barring AI companies from training on copyrighted material could "severely hamper" technological progress, framing model training as a transformative use similar to journalism. This filing marks a pivotal moment where the federal government explicitly aligns with tech giants on the interpretation of fair use for LLM training, potentially influencing Judge Sidney Stein’s upcoming rulings.
Seattle Times and Newsday Sue for Model Destruction
On September 5, 2026, The Seattle Times and Newsday filed a lawsuit against OpenAI and Microsoft in Manhattan federal court. The plaintiffs cited 1,375 registered works and are seeking not only damages but also the destruction of the specific AI models and training data derived from their copyrighted content. This case adds to the growing volume of publisher litigation, with over 140 copyright lawsuits now active against AI companies in the US as of September 2026, up from 45 in June 2025.
Authors Guild Seeks Summary Judgment Against OpenAI
The Authors Guild has requested a decision without a full trial (summary judgment) in its class action against OpenAI and Microsoft. Filed around September 10, 2026, the motion argues that the use of pirated books from the LibGen dataset for training constitutes clear infringement, challenging the "fair use" defense based on market harm. This move aims to accelerate a ruling on the legality of using pirated data, a key distinction from cases involving legally purchased or licensed datasets.
German Court Rules Against OpenAI in GEMA Case
In a development contrasting with US trends, the Munich Regional Court ruled that OpenAI violated copyright laws by reproducing song lyrics without license in ChatGPT outputs. The case, brought by German collecting society GEMA, found that the model's ability to generate near-verbatim excerpts of protected lyrics constituted infringement. This ruling underscores the stricter liability standards in the EU regarding output generation, even if the training process itself is under different scrutiny.
Local view
Germany: German media outlets like FAZ and Netz-Trends are closely covering the aggressive stance of US publishers like the Seattle Times, noting the demand for the "destruction" of trained models. The local legal community is watching these US developments while simultaneously dealing with domestic precedents like the GEMA ruling, which establishes a high bar for output compliance.
Japan: Japanese business media reported on the US DOJ's support for OpenAI's fair use claim, viewing it as a critical signal for global AI policy. The coverage highlights the tension between Japan's "Article 30-4" copyright exception, which allows broad text and data mining, and the increasingly litigious environment in the US.
Context & numbers
- Lawsuit Volume: US copyright lawsuits against AI companies have reached 140 cases by September 2026, a sharp increase from 45 cases in June 2025.
- Licensing Deals: Despite the litigation, licensing continues. OpenAI has signed approximately two dozen publisher deals, with the largest disclosed deal being $250 million over five years with News Corp.
- Anthropic Settlement: The Authors Guild continues to manage the distribution of notices for the $1.5 billion Anthropic settlement, with claim summaries being sent out to authors this week.
On the radar
- Judge Stein's Ruling: All eyes remain on U.S. District Judge Sidney Stein in Manhattan, who is expected to issue a significant ruling on whether LLM training qualifies as fair use in the NYT v. OpenAI case. No appellate court has yet issued a substantive ruling on this core issue.
- UK Policy Shift: Following reports that the UK government has moved away from a preferred position on a broad text and data mining exception, further consultations or legislative updates are anticipated as the government prioritizes "licensing-first" approaches.
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
