It’s been a while since I have written about AI and copyright, although I have been presenting about it quite a lot. I wrote a paper entitled “Do Androids Dream of Electric Copyright?” in 2017, and then a shorter version for WIPO Magazine (it’s sobering how many more people cite the shorter version than the academic paper, but I digress).
TL;DR version of the paper, I go through the way in which various jurisdictions deal with works generated by artificial intelligence, and I conclude that the best option is how the UK handles them. Section 9(3) of the UK Copyright, Designs and Patents Act 1988 (CDPA) states that the author of a computer generated work is the person who made the arrangements necessary for the work to be created. I believe that for the time being this is the best option out there.
Interestingly, there is an ongoing debate on this subject, as various scholars have made their positions clear on the subject, with three clear positions emerging:
- The default position, and the one that seems to have more scholarly support, is that there is no copyright because there cannot be originality in a work that has not been created by a human, and without originality there is no copyright.
As mentioned, I advocate for a legislative declarative approach similar to that in UK law, where the question of authorship is solved by simply allocating authorship to whoever made arrangements for the work to come into being.
A third option, which emerged mostly from a series Japanese strategy policy papers on the subject, would be to bypass copyright altogether and provide a new type of right for AI creators, this sui generis right could be something akin to the European database right, and it would be something that would protect investment. These proposals haven’t been implemented.
The prevalent strategy everywhere but in a few countries that have implemented the UK system is to keep these works in the public domain as there cannot be copyright protection. There cannot be a work without a human author. Moreover, the requirement for originality in Europe is that a work must be the intellectual creation of the author, and if the author did nothing but press a button and set some parameters, then there is no intellectual creation.
While persuasive, there are various ways in which we can maintain the existing originality requirements, and still have some sort of protection for AI generated works. Firstly, the task of generating a work using AI is often not just a matter of pressing a button and letting the AI do all the work, someone has to program and teach the computer to compose music, write, or paint, and this is a process that is both lengthy and full of intellectual creativity. The makers of works of art such as The Next Rembrandt engaged in lengthy process, which could have enough “intellectual creation”. If you visit a website like Jukedeck, press a button, and obtain a song, then by all means, we could argue that work will not have protection. But more sophisticated training of an AI could potentially be considered as carrying enough intellectual creation from the programmers.
Secondly, the current system relies on the concept of originality that is very human centric, be it the requirement of intellectual creation, or the requirement of creativity in the US. The idea behind this is that the human creative spark itself is what imbues a work with protection. But we are perfectly happy allowing legal persons to be authors and copyright owners, granted, with the understanding that the works are created by humans, but why not continue having another legal fiction only for AI works?
Thirdly, originality used to subsist if the author had exercised enough skill, labour and judgement to warrant copyright protection. Why not go back to a similar system that rewards a “sweat of the brow” approach? I know that this is passée, and that the approach of giving copyright as a reward for effort is no longer considered viable, but it might be worth reviewing the merits of recognising the amount of effort and investment that goes into the creation of some of these works.
Finally, there are a number of practical problems with allowing increasing numbers of AI works to co-exist with human works. It is possible that public domain AI works will result in some creators to go out of business, as they cannot compete with free works. Stock photography, jingles, music, journalistic pieces, all of these could be affected by increasingly sophisticated AI. One could argue that the world will not miss armies of jingle composers, but many musicians make a living from this type of work.
Moreover, I do not think that people have really thought about the practical implications for copyright. Copyright has no registration, so a work is assumed to be protected if it meets the requirements. This assumption is often recognised by everyone, and it is not usually tested in court unless there is a conflict. The increasing sophistication of AI works will mean that there will be growing doubt as to the legitimate origin of works. Is this music created by a human or by an AI? How could you tell?
Concluding, I think that this topic will continue to generate a lot of interest. I believe that there is a good case to be made as to either provide protection through a norm similar to section 9(3) CDPA, or to go for a sui generis right. I strongly believe that commercial pressure will push us in one of these directions.