The Litigious Arrival of Artificial Intelligence in Light of Copyright Law

The emergence of generative artificial intelligence has been faster and more disruptive than any recent technological development. Unlike social media, wearables, or augmented reality, AI has fundamentally reshaped professional, academic, and social dynamics.

From predictive text to advanced language models such as GPT, PaLM, and Claude, embedded in tools like ChatGPT, Copilot, and Gemini, AI has become ubiquitous for users interacting with digital technologies.

From Skepticism to Structural Adoption

AI adoption has followed a gradual path. Initial skepticism centered on trust: how could systems that fabricate norms, concepts, or citations be reliable? This was followed by an exploratory phase, where AI proved useful for specific tasks such as drafting texts, translations, or image generation.

Today, AI is entering a structural transformation phase, reshaping how work is performed, communication takes place, and problems are solved.

Training Data and Copyright Concerns

As AI becomes normalized, less attention has been paid to the origin of the content used to train these models. Key legal questions include:

• What data sources were used for training?

• Were copyrighted works included?

• Do AI-generated outputs pose infringement risks?

• Can “regurgitation” of protected content occur?

These issues have fueled extensive litigation, particularly in the United States. As of June 2025, more than 45 lawsuits had been filed alleging copyright infringement related to AI training, involving plaintiffs such as authors, publishers, The New York Times, and Disney.

Early Judicial Responses in the United States

In June 2025, federal courts issued landmark decisions in Bartz v. Anthropic and Kadrey v. Meta. Both rulings suggested that using copyrighted books for AI training may qualify as fair use, though their reasoning differed.

One emphasized the transformative nature of AI training, likening it to human learning, while the other raised concerns about market harm and reduced incentives for human creativity. Judges also diverged on whether training AI using illegally sourced works could be justified.

Potential Impact Beyond the United States

Although these rulings are not binding outside the U.S., and fair use has no direct equivalent in Colombian or Andean copyright law, their influence may extend internationally.

They could shape interpretations of copyright limitations and exceptions, especially considering the openness shown by the Andean Court of Justice in cases such as Interpretation 135-IP-2020, which acknowledged the applicability of foreign concepts like incidental use.

Conclusion

The legal debate surrounding artificial intelligence and copyright is only beginning. Judicial interpretations will continue to evolve until higher courts or legislators establish clear frameworks.

In the meantime, understanding these developments and anticipating their impact is crucial. The key challenge lies in striking a balance between technological innovation and the effective protection of creators’ rights.

WRITTEN BY: CRISTIAN SARMIENTO

CONTACT US