It is common for businesses to promote their use of artificial intelligence (AI) as a tool to better their efficiency, effectiveness, and quality. There is no doubt that modern developments in technology are revolutionising the creation of products and expression of ideas. Thus far, English copyright law has been able to accommodate these changes, but it is uncertain whether it will be able to withstand the next wave of development.
In 2016, 20th Century Fox teamed up with IBM and its AI, Watson, to produce a film trailer for Morgan. The team trained Watson by showing the system scenes from ‘100 horror movies’. When they believed the AI had a suitable understanding of the genre of suspense and horror, they played all of the movie footage to it and asked it to select 10 ‘moments’ to create a 6 minute trailer for the film. The human involvement in the production of the trailer was minimal. It was the AI alone that selected the clips to be used 1.
In such circumstances, it is arguable whether the humans ‘used’ the AI to create the trailer, or whether they set up the AI and it created the trailer itself. Andres Guadamuz notes that when human interaction is minimal or non-existent, there are two ways in which copyright law can deal with the works that are produced2.
The first option is finding that there is no author of the work and thus concluding that there can be no copyright over it. Legal systems which hold that an author of a work must be a human being, such as those of Spain and Germany3, would come to this conclusion.
This is the approach taken in Singapore. Although it was a case about a corporation seeking copyright, in July 2011, Singapore’s Court of Appeal laid out key principles of its copyright law in Asia Pacific Publishing Pte Ltd v Pioneers & Leaders (Publishers) Pte.The Court of Appeal held that a work cannot be copyrighted unless it has an author and an author may only be an individual human being; a company cannot even qualify as an author. It stated that “not infrequently, in cases involving a high degree of automation, there will be no original work produced for the simple reason that there are no identifiable human authors”4. The Court of Appeal gave many reasons for its findings, but the most significant was an issue of public policy. The court noted that by awarding copyright to a human being, the length of that copyright was inherently limited by the length of the human’s life5. If copyright were to be awarded to in this case, a ‘non-living’ corporate entity, or in general, any non-living entity, such as AI, this would allow copyright protection in perpetuity. Such a result would stifle creative innovation and development.
Australia takes a similar position. In Acohs Pty Ltd v Ucorp Pty Ltd [2012] FCAFC 166, the Federal Court of Australia held that, whatever originality might have been involved in the activities undertaken by the programmers in writing the underlying computer programs, those activities were separate and distinct from the activities involved in creating each product, which were performed by technology. The intervention of technology meant that the work could not be copyrighted.
The second option by which to address this issue is the approach taken by the UK; the recognition of the programmer as author. The Copyright, Designs and Patents Act 1988 (‘CDPA’), a statute central to copyright law in the UK, lays out the following provisions:
Section 9(3):“In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.”
Section 178:“computer-generated”, in relation to a work, means that the work is generated by computer in circumstances such that there is no human author of the work7;
In applying these provisions to the movie trailer scenario above, it is clear that the programmers of the AI would be given copyright over the work produced. Such an approach also resolves the policy issues noted by the Singaporean Court of Appeal. A human, not a non-human entity, is being granted authorship of the product.
An issue with this approach, however, is that the originality of the work is not being honoured by the designation of authorship. Section 1(1) of the CDPA acknowledges that copyright is a property right which subsists in: (a) original literary, dramatic, musical or artistic works, (b) sound recordings, films or broadcasts, and (c) the typographical arrangement of published editions8. In awarding authorship to the programmer of AI, the work that is being protected is the work of programming technology. The AI itself occasioned the creative spark which led to the product.
The current state of UK law is sufficiently able to deal with the use of AI through the provisions mentioned above. It is, however, unsuitable to deal with potential future developments. Currently, AI is sufficiently under developed such that a great deal of human interaction is still required for a product to arise. However, as AI becomes more self-sufficient, human interaction will become less invasive and further removed from the product. In such a case, is it acceptable to award authorship to the human now barely connected to the work? Would such recognition undermine the purpose and foundation of copyright law?
If AI were to become fully autonomous, to the extent that no human would have to make arrangements for the creation of the work, under both options, no copyright could exist9. This situation carries the same disadvantages as the first option. Without copyright law, technological development would become a free-for-all and the system of intellectual honesty would disappear overnight.
One suggestion is that in order to give AI-derived literary works IP protection, it may be necessary to legislate for a new sui generis right that expressly protects works created by AI10. In stark contrast, Andres Guadamuz argues that granting copyright to the person who made the operation of AI possible is the most sensible approach. He notes that it will keep companies investing in the technology, safe in the knowledge that they will get a return on their investment11.
It is clear that UK law contains sufficient protection for the current state of technological development. Guadamuz’s argument is compelling, as the current policy implications of stalling technological development are monumental. Recognising the importance of the policy issues and the need to avoid the consequences of the abolition of copyright law entirely, it may be necessary to continue to recognise any human in contact with the AI, no matter how minimal or far removed from the actual production of the work.
The law in the UK, which prevents authorship vesting in the AI itself, cannot currently be considered unfair. It is only if we attribute human status to the AI, and thus accept that it is the AI itself which should be vested with authorship, that this law becomes unjust and misrepresentative of the true assertion of intellectual effort.
