The European Union, the United States and Creative Works in the Age of AI

Analysis of the United States and European Union's copyright laws as it pertains to Artificial Intelligence

---

Published Februrary 15th, 2026

Written by Evan Randle

The very nature of copyright and intellectual property law suggests constant revision and evolution. How lawyers protect the rights of creators across disciplines is, in some way, dependent on new technologies and how conceptions of ownership may change as a result of these technologies. The boom of Artificial Intelligence over the past few years has proven incredibly relevant, even in such a flexible and ever-changing field. Various countries and lawyers have struggled to make sense of copyright and IP law since. Though permanent solutions appear challenging, initial laws have been enacted in the most influential countries. As proof of a lack of universal solution, an analysis of the United States and the European Union’s copyright law vis-a-vis AI reveals relevant points of divergence, particularly in the sphere of the creative industry.

Regulators of the E.U. attempt a balancing act between fostering AI growth and protecting the rights of artists. A major piece of legislation across the E.U. is the Artificial Intelligence Act, enacted in 2024, according to a Bruegel article. In more recent months, the European Commission has published guidelines for the implementation and usage of AI, such as the July 2025 Code of Practice. The CoP covers multiple issues, the most contested of which is how to comply with copyright while training AI. This section restricts models from producing copyright-protected content in their outputs and from training on unlawfully accessible data. In effect, this means that pirated books, video games and other media cannot form the basis of AI training. Furthermore, the European Union allows right holders to object to the use of their creative property in AI training, according to a Rand article from 2024. This all results from delicate balances that the E.U. must play to satisfy creatives while remaining globally competitive in the increasingly relevant AI industry. Bruegel reports that the Artificial Intelligence Act and the following CoP hope to be transparent about what material trains AI while handling creatives who have opted out of their work being used to train AI models. There is a chance, however, that this may backfire — the rising costs placed on AI companies as a result of having to comply with regulations could lead to individual companies designing their own IP laws to satisfy executives. Rand conjectures that AI companies could simply opt not to enter the EU market at all. Overall, the EU provides more opportunities for the preservation of copyright at the cost of potential business losses.

In contrast, the United States encourages innovation and AI advancement first and foremost, potentially undermining the creative rights of artists across the country. Compared to countries like Japan, Singapore and groups like the E.U., Rand reports that the United States does not place restrictions on Text Data Mining training on copyrighted work. In addition, the U.S. does not currently require AI companies to be transparent nor fine them for refusing to state where they derive data for training. This has created significant outcry within artist communities, who argue that unchecked AI training could not only lead to plagiarism but also to an attempt to replace their work with that generated by models trained on their work. An implicit rule in U.S. copyright law, as it pertains to creative works, is that AI can only learn from artists, not attempt to replicate them. A violation of this rule could lead to infringement claims, though — unlike in other regions of the world, such as Europe — these claims are not guaranteed to succeed by any governmental efforts to regulate AI. Most of the world seems, so far, to have followed in the U.S.’s footsteps, as many countries encourage the use of copyrighted works to train AI models, the Emory Law Journal writes. This primarily results from countries hoping to be globally competitive in the AI sphere, as well as larger ambiguities surrounding the relationship between ideas and expression in copyright law.

There seems to be no global consensus on the role of copyright, intellectual property, and AI, other than the seemingly omnipresent worry of countries that they will lag in the race to master artificial intelligence. What may follow from this competition, however, is a lack of true global understanding of how creative works can maintain their legitimacy when algorithms could replicate their substance. As the efforts of the European Union have shown, the development of AI models and the protection of creative works are not mutually exclusive — they merely require nations willing to commit to the continued presence of artistic integrity.

---