Source: Vorys

Vorys provides a legal analysis of the central question in the ongoing litigation between copyright holders and AI developers: whether training generative AI models on vast amounts of copyrighted material constitutes fair use. The publication examines recent court rulings that are beginning to shape this critical area of intellectual property law. The core of the dispute revolves around the four factors of the fair use doctrine: the purpose and character of the use, the nature of the copyrighted work, the amount of the portion used, and the effect of the use on the potential market for the original work. AI developers argue that using works for training is a transformative and non-commercial "intermediate" use, intended to create entirely new outputs rather than to substitute the original works.

However, as the article by Jake Evans and a|Top Caputo points out, copyright owners contend that this training process involves unauthorized copying on a massive scale and that the resulting AI-generated outputs can directly compete with and devalue their original creations. Recent judicial decisions have shown a nuanced approach. While some early rulings have been skeptical of broad infringement claims at the motion-to-dismiss stage, particularly where direct evidence of substantially similar output is lacking, no court has definitively ruled that AI training is, as a whole, fair use. The analysis suggests that the "transformative use" argument is pivotal and that future decisions will likely depend heavily on the specific facts of each case, especially the degree to which AI-generated content mimics or supplants the market for the copyrighted training data.