AI and IP (artificial intelligence and intellectual property)

Although artificial intelligence (AI) has not featured prominently in most people’s lives up until now, examples of how it can be used have been becoming more visible recently. Take, for example, GPT-3, a programme that writes texts autonomously based on input. Sometimes the results are astonishing; for example, a recent tech-podcast was written by this programme.

Programmes like Dall E 2 and Midjourney, which create images, such as photos are also constantly improving. For example, if you give Dall E 2 verbal instructions, such as ‘two trees, Van Gogh style and a sunset’, it will get to work on creating new images by drawing on a database of millions of photos.

Obviously, such developments are very handy, especially for content makers. But are works of this type protected on the grounds of intellectual property? If so, what law should they fall under? The most obvious choice is probably copyright law, as original works of art, literature, photographs etc. are protected under copyright. An important condition, however, is that a work must show creative choices. How could copyright be applicable if the creative choices were made mainly by a computer programme?

It is a different matter, however, if the programme is only a tool and the artist’s contribution is clearly manifest. For example, it was possible to register a comic book created using AI as a copyright-protected work in America. There is something to be said for this, seeing that the artist had written the story and devised the headings and layout. In short, the creative choices were made by the maker, not by a computer.