US court rejects AI fair use defense Thomson Reuters wins copyright

Lucas Bennett 09/30/2026

AI Training Data Fight: Appeals Court Upholds AI Copying of Protected Material as Legitimate

The legal battle between artificial intelligence and copyright is entering a new phase, with a significant decision from the federal appeals court setting a major precedent for how AI models can be trained. The court has upheld a ruling that allows AI companies to copy protected material to build the vast datasets required for training their systems, determining that this practice does not constitute unfair use.

This decision is a landmark moment in the ongoing debate over intellectual property rights in the digital age. For years, the core legal question has been whether the ingestion of copyrighted works—music, text, images, and code—into large language models and generative AI systems crosses the line into infringement. Now, the court has firmly sided with the developers, acknowledging the necessity of this data usage for technological advancement.

The ruling essentially validates the practice of using copyrighted material as training data, suggesting that the transformative nature of AI development permits this copying under existing fair use principles. This finding is crucial because it addresses the foundational mechanism by which modern AI systems learn and evolve.

Previously, the complexity of applying traditional copyright law to the scale and speed of AI development left many legal experts and creators in limbo. This appeals court decision provides a clear, albeit complex, framework for navigating the intersection of technology and creative law.

By upholding the defense, the court acknowledges that the process of machine learning requires access to existing works, reframing the debate from simple infringement to the functional necessity of data acquisition for innovation. This shifts the focus toward how copyright law should adapt to the digital reality of automated creation.

The implications extend far beyond the immediate case. This decision signals that while creators maintain rights over their work, the pace of technological innovation must be balanced against the utility of leveraging existing culture for future progress. It opens up new avenues for developers to build increasingly sophisticated AI tools without facing immediate legal barriers regarding their training sets.

Ultimately, this legal affirmation underscores a fundamental tension: the need to protect creators’ livelihoods versus the drive for technological advancement. As AI continues to evolve, the courts are now tasked with ensuring that the engines of innovation can run smoothly while still respecting the foundational principles of authorship and ownership.