No authors at all, or too many to count? – AI and IP

Published 29 July 2019

What is AI, and how does it create original works?

Artificial intelligence (or “AI”) describes a simulation of human intelligence, produced by a machine. It allows computers to reason, process information and, more recently, has been disrupting artistic disciplines by creating “original” works by processing, copying and reproducing existing works into new pieces.

From Chinese tech firm Huawei’s completion of Schubert’s unfinished symphony No. 8 to Dutch banking group ING’s The Next Rembrandt project, firms are increasingly recognising and using the power of AI to create high quality, original works using data collected from the priceless work of renowned artists. The projects have seen raw data such as existing musical scores by Schubert and intricate 3D digital scans of Rembrandt paintings input into AI systems, which are then able to produce original works in the style of the original artist. The unfinished symphony was finished with the assistance of data collected from the 1st and 2nd movements of Schubert’s Symphony No 8 and The Next Rembrandt was 3D printed to mimic Rembrandt’s preferred portrait style and brush stroke techniques – all learned and reproduced by the power of AI.

Can AI-enabled work attract IP protection, and who is the ultimate author?

There is very little doubt that content generated by AI is, and should be, capable of attracting IP protection, although questions are raised as to who owns the IP. These fundamental questions begin with the originality and authorship of such works. The test for originality requires works to be of the “author's own intellectual creation". This question cannot be answered without first turning ones attention to who the author is; and this is where AI provides a unique uncertainty, as there are many potential authors, one of which could even be a machine incapable of independent thought.
The Copyright, Designs and Patents Act 1988 (“CDPA”) defines a computer-generated work as one which has been produced “in circumstances such that there is no human author of the work” (s 178). At present, most AI-enabled works require human involvement, for example:

1. the original artist who created the raw data;
2. the creative directors who input data into an AI system (Huawei or ING via their employees) or even, in the Huawei example, its consumers who have access to the AI software via their handheld mobile devices;
3. the software developers who created the algorithms; or
4. any other natural living persons who completed or corrected the final works.

Where there is human involvement, the CDPA provides that, the author of a computer generated work “shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken”. (s 9(3)). It appears, therefore, that no protection could be afforded in circumstances where AI created works are fully autonomous.

All of the above parties arguably have made arrangements that are necessary for the creation of a work, so a clear author cannot be determined. Academic commentators appear to believe that the wording of the legislation seems to favour the rights of any natural living person who completed or corrected the final work above other possible contenders. This is logical as the CDPA was drafted pre-1988, when artificial intelligence, as it is today, had almost certainly not been contemplated. Therefore, the CDPA is likely to be referring to computer-generated works where computers are used as a mere tool, for example, a camera being used to take a photograph, rather than an intelligent learning system capable of producing its own work.

In all of this confusion, three schools of thought have emerged. There are those who:

1. favour the rights of one or more of the above candidates (most commonly candidate 2 due to them being perceived to have the most involvement in the creative process);
2. reject that AI generated works give rise to any protection at all, or would prefer AI-enabled works to be limited to deserving of economic rights only; and
3. believe matters can only be fairly determined on a case by case basis.

What happens now?

The courts are yet to give a decisive ruling as to the authorship of AI-enabled works, which is guaranteed to be the subject of dispute in the future. As AI software becomes more advanced, and its abilities more sophisticated, it will be imperative for legislators and the courts to provide clarity over these questions or businesses may face the situation where fully autonomous AI-enabled works may attract no protection at all.