AI Copyright: Training-Data Lawsuits and Licences — 2026-10-16
The legal landscape for AI training data has sharpened significantly this week, with the Third Circuit Court of Appeals ruling against a fair use defense in *Thomson Reuters v. Ross Intelligence*, setting a precedent that could severely impact music and publisher lawsuits. Meanwhile, the European Commission has launched a targeted consultation on generative AI copyright compliance, signaling potential regulatory shifts in the EU. In Japan, new transparency requirements for AI training data are coming into focus as local media reports on upcoming rule changes.
AI Copyright: Training-Data Lawsuits and Licences — 2026-10-16
Top developments
US Appeals Court Rejects Fair Use for AI Training in Thomson Reuters v. Ross
On October 15, 2026, reporting highlighted a pivotal US federal appeals court ruling from late September that upheld a lower court decision against AI legal research company Ross Intelligence. The court determined that copying protected material to build training data did not constitute "fair use," a decision backed by industry groups like the RIAA and NMPA. This ruling is critical for pending cases against music AI companies like Suno, as it establishes that the acquisition of pirated or unlicensed data for model building is not shielded by fair use doctrines, potentially exposing AI developers to significant liability for their training datasets.
USA Today Sues OpenAI for $250 Million Over Training Data
In a significant escalation of publisher-AI litigation, USA Today Co. filed a lawsuit against OpenAI, alleging unauthorized use of hundreds of thousands of copyrighted articles for training its AI models. The complaint seeks damages exceeding $250 million and an injunction to stop further use of their content. This case adds to the growing list of media giants challenging the legality of "fair use" defenses in the context of large-scale text mining, arguing that such use harms their ability to license content directly.
EU Commission Launches Consultation on Generative AI Copyright
On September 29, 2026, the European Commission opened a targeted consultation to gather stakeholder views on how generative AI affects copyright protection. The consultation focuses on four key topics, including compliance with existing rules for AI training data. This move signals Brussels' intent to potentially tighten the regulatory framework around AI development, moving beyond the current "opt-out" mechanisms toward more explicit requirements for transparency and remuneration for rights holders.
National Review Analyzes Key Questions in AI Copyright Cases
Recent legal analyses, including a piece published three days ago by National Review, emphasize that courts are increasingly distinguishing between how AI companies acquire copyrighted material, how models use it, and whether outputs infringe. The analysis notes that rulings involving Anthropic and Ross Intelligence are clarifying that "fair use" is not a blanket shield, particularly when pirated books or competing products are involved. These distinctions are becoming central to how courts evaluate settlement offers and licensing deals.
Local view
Japan: Mainichi Shimbun Reports on New Transparency Rules Japanese media outlet Mainichi Shimbun reported on October 15, 2026, about new rules requiring AI operators to provide certain explanations regarding the data used to train their models. This aligns with ongoing discussions around Article 30-4 of Japan’s Copyright Act, which permits certain uses for non-enjoyment purposes but faces scrutiny as AI capabilities grow. The report highlights a shift towards mandatory disclosure, aiming to balance innovation with creator rights.
Context & numbers
- Lawsuit Volume: A public tracker lists 248 AI lawsuits against 76 companies as of October 2026, with OpenAI facing 55 cases and Google facing 25. The Third Circuit Ross ruling is cited as a key precedent in these proceedings.
- Settlement Precedent: Anthropic previously agreed to a $1.5 billion copyright settlement regarding pirated books, which received final approval on July 20, 2026. This figure remains a benchmark for potential damages in similar cases involving large-scale dataset piracy.
- Licensing Values: Major licensing deals continue to set market rates. News Corp’s deal with OpenAI was valued at up to $250 million over five years, while Apple reportedly approached publishers with offers worth at least $50 million for news archives. These figures provide context for the damages sought in new lawsuits like USA Today’s.
On the radar
- NYT v. OpenAI Discovery: The New York Times v. OpenAI & Microsoft case continues to progress, with Judge Stein’s January 2026 order affirming earlier rulings. Observers are watching for any new discovery disputes or summary judgment motions that could clarify the scope of permissible training data usage under US law.
- EU Consultation Deadlines: Stakeholders in the European Union have until the end of the consultation period to submit feedback on the four key topics raised by the Commission. Responses will likely influence whether the EU introduces stricter transparency obligations or remuneration schemes for AI developers.
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
