AI Copyright: Training-Data Lawsuits and Licences — 2026-09-11
The US Department of Justice has formally filed a statement of interest supporting OpenAI and Microsoft, arguing that barring AI training on copyrighted material would "severely hamper" technological progress. Meanwhile, the Seattle Times and Newsday have joined the legal fray with a new lawsuit against OpenAI and Microsoft, demanding the destruction of models trained on their content.
AI Copyright: Training-Data Lawsuits and Licences — 2026-09-11
Top developments
DOJ Intervenes in NYT v. OpenAI Fair Use Battle
On September 4, 2026, the US Department of Justice filed a statement of interest in the New York Times v. OpenAI case, siding with the AI developers. The DOJ argued that training models on copyrighted works constitutes "transformative use" under fair use doctrine, warning that a contrary ruling could stifle innovation. This intervention comes as Judge Sidney Stein in Manhattan prepares to rule on whether LLM training qualifies as fair use, a decision that could set a precedent for the entire generative AI industry.
Seattle Times and Newsday File New Suit Against OpenAI and Microsoft
In a significant escalation, the Seattle Times and Newsday filed a class-action lawsuit in Manhattan federal court on September 5, 2026, against nine OpenAI entities and Microsoft. The plaintiffs allege copyright infringement involving 1,375 registered works and are seeking not only damages but also an injunction to prevent future training on their content and the destruction of existing models trained on their data. The suit highlights growing publisher frustration with alleged verbatim reproduction of articles by ChatGPT and Copilot.
Authors Guild Pushes for Summary Judgment in Anthropic Settlement
The Authors Guild is urging the court to resolve the Bartz v. Anthropic class action without a full trial, focusing on the dispute over whether training data from LibGen was used illegally. On September 4, 2026, settlement administrators began sending notices to claimants detailing their individual claims and any overlapping filings for specific titles. This move follows the court's earlier recognition that while copying entire books was "reasonably necessary," the market effect factor favored Anthropic because training data is not directly shown to users.
Local view
In Germany, the debate over AI training legality intensified this week following a landmark first-instance ruling by the Munich Regional Court (LG München I) in favor of GEMA against Suno. The court determined that using copyrighted song lyrics to train AI without a license violates German copyright law, ordering damages though the final amount remains open. This decision reinforces the German stance that unauthorized training data usage constitutes infringement, contrasting with the more permissive fair use arguments currently being tested in US courts.
Context & numbers
The volume of litigation continues to surge, with US copyright lawsuits against AI companies reaching approximately 140 cases by September 2026, up from 45 in June 2025. In terms of licensing, OpenAI’s reported $250 million five-year deal with News Corp remains the largest disclosed content licensing agreement to date, serving as a benchmark for ongoing negotiations.
On the radar
- Judge Stein's Ruling: All eyes are on U.S. District Judge Sidney Stein in Manhattan, who must now decide if LLM training on news articles qualifies as fair use, potentially reshaping AI economics.
- Sony/Warner vs. Anthropic: Legal proceedings continue regarding allegations that Anthropic trained Claude models on unlicensed lyrics owned by Sony and Warner Music Group.
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
