Copyright: does AI change everything?

This has changed with the advent of industrial printing presses, photography, phonography, radio, television, photocopiers and the internet, as these inventions have made it possible to reproduce or distribute works cheaply and on a massive scale. The emergence of new reproduction and distribution technologies has also triggered questions about how to protect new art forms (such as photography) and new service providers, such as record producers and broadcasters. Copyright law has continued to evolve as a result of these technological upheavals. Hence, the Swiss Copyright Act (CopA), makes it clear that authors have an exclusive right to make their works available via the internet (Art. 10 para. 2c CopA). It’s therefore unlawful to upload a photograph to a website or social media without the photographer’s consent. Providers of internet hosting services are also obliged to prevent works or other protected subject-matter from being made available via their services a second time. This means they have to ensure that copyright-infringing content that has been removed following a notice from the rights holder is not illegally uploaded again (the ‘stay down’ obligation, Art. 39d CopA).

Unlike earlier technologies, such as photocopiers, generative AI is not primarily used to reproduce and disseminate existing works. Instead, it can generate new content itself, such as text, images or music. In this way, AI takes over part of the creative process and (partly) replaces humans as creators. It also calls into question the distinction between the terms ‘original’ and ‘copy’. Unlike with photocopying, AI does not normally produce an exact replica; however, its creations may bear a very strong resemblance to existing works. Its ability to do so stems from the fact that it’s been trained using copyright-protected works, among other things. AI thus calls copyright into question at a very fundamental level: Who can actually be an author? How much human involvement is required for us to speak of a ‘work’ in the legal sense? And how can innovation be encouraged without putting creative professionals at a disadvantage? What makes the regulation of AI unique is that, for the first time, technology is producing output that can compete with works created by authors. This means that debate is no longer limited to the traditional copyright question about the conditions under which existing works may be used. There is also growing discussion about how human creativity should be protected when it forms the basis for systems whose products are offered on the same markets as the works of creators. The aim of regulating AI is thus to strike a balance that goes beyond the original conflict between the protection of existing works and interest in their use: indeed, regulations should help clarify the conditions under which human creativity may form the basis of artificial creations.

Franziska Raaflaub