Artificial Intelligence Regulation: Addressing the Copyright Ownership Dilemma of Machine Generated Works should be a First Priority

The concept that machines can produce human-like outputs and interactions has been a matter of science fiction for decades. However, thanks to various technological advancements and access to more computing power, we now have a reality where machines can artificially perceive, think, understand, learn, produce and interact without explicit human programming. Put simply, machines are becoming capable of producing works without human intervention and crucially, in a way that was not necessarily directly predicted or coded by the creator.

This new reality brings about various legal and philosophical challenges and dilemmas: the whole legal system was after all built on the presumption that there is some form of human intervention. In this respect, Malta has recently taken the commendable step of establishing a Taskforce entrusted with devising a national strategy for the purposes of some form of regulatory intervention in respect of regulating Artificial Intelligence (“AI”).

Malta’s tapping into the regulation of AI had been predicted for some time and the Maltese Government made it clear that the legal framework regulating distributed ledger technologies can be in some way extended to cover AI and eventually also internet of things technologies. The subject has also received some public debate amongst practitioners and scholars, typically centering around the need for AI ethics, standards, safeguards and liability for decisions taken by an AI. Somewhat controversially, the Government of Malta also announced a project being undertaken in collaboration with the company SingularityNET which may look into some form of citizenship or other rights being granted to AI technologies.  

Against this backdrop, it should be highlighted that amongst researchers there seems to be a general agreement that AI is not as not advanced or smart as it is typically portrayed to be. In fact, the technology can be described to be still in its embryonic stages and a far cry from its depiction in films and science fiction where sentient beings are a reality. However, whilst an AI induced doomsday is still a matter for science fiction, we already have a proliferation of AI in the creative industries, for example, in 2016 “Daddy’s Car” was published – the first song ever authored by an AI. In fact, it is already common that AI is some way used in photos, films, music, content creation, graphic design, news and games to mention a few creative industries. We therefore already have a considerable existing body of intangible works created by machines without any direct human involvement. An elephant in the room can be said to exist under Maltese law in this respect, one which the upcoming regulation on AI must tackle as a priority: do machine-generated intangible works qualify for protection? If yes, who is the rightful owner?

The point of departure for any new legal regime should be analyzing whether the existing laws already cater for the matter at hand. In this respect, it should be noted that under the current legal system, intellectual creations are protected through intellectual property rights (“IPRs”). At law, IPRs can take various forms, some of which are registrable and others which arise without the need of registration. Every type of IPR protects a different form of intellectual effort. Patents, for example, protect inventions while trademarks protect distinctive indications of origin. All IPRs will need to be given careful consideration in the context of an overall national strategy to properly regulate AI and machine-generated works, however, given the state of the art achieved in AI technology, this article will focus exclusively on copyright as the only IPR which protects the external manifestation of artistic intellectual efforts and outputs.

The Copyright Act (Chapter 415 of the Laws of Malta) does not directly prohibit copyright conferment to works produced by AI systems but it seems to clearly point towards the direction that the author of a work which is eligible for copyright must necessarily be a natural person. Indeed, the term “author” under the Copyright Act is defined as meaning a “natural person or group of natural persons who created the work eligible for copyright”. Essentially, this could result in a situation where we have a considerable number of intellectual creations which fail to be eligible for copyright simply because they were created by a machine.

This position could be interpreted to reflect the position being inferred at EU level, given that in Infopaq International A/S v Danske Dagbaldes Forening (C-508) the Court of Justice of European Union (“CJEU”) ruled that copyright applies only to original works, and that originality must reflect “the author’s own intellectual creation”. By inference, this can be taken to mean that direct human intervention is necessary for copyright to be conferred. However, it must be highlighted that this case concerned the rights of reproduction of copyrighted work under Directive 2001/29 and the conditions for exemption therefrom, in the context of a media-monitoring service which employed a data capture process using search words to produce summaries of articles for clients, and thus was not directly addressing the same point.

If one had to look at the practice which is developing in foreign jurisdictions, one immediately notices that there is a conflicting approach on this point which can be particularly problematic for copyright, given that its cross-border usefulness depends on automatic foreign recognition of works produced in other countries through conventions such as the Berne Convention.

Countries such as the United Kingdom (generally, the source of Maltese IP laws and case-law), Ireland, Canada, New Zealand and India took the approach of recognizing that machine-generated works are eligible for copyright protection, and by extension, bestow the copyright to the programmer. Indeed, article 9(3) of the United Kingdom’s Copyright, Designs and Patens Act (1998) provides:

“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.”

Other countries such as the United States and Australia seem to have taken the very opposite approach. In fact, in Acohs Pty Ltd v Ucorp Pty Ltd (2010) FCA 577, the Australian did not recognize copyright protection for a work generated through computer intervention simply because it was not produced by a human being.

Regulating ownership of machine-driven creativity and intangible works will have various far-reaching commercial and legal implications and it is submitted that this will be key to the success of any legal framework regulating AI. This will be no easy task to address in a manner which confers legal certainty especially as this point could have some spillover effects on other legal notions such as liability. However, having a situation where machine-generated works fail to qualify for copyright protection simply because they were created by an AI should be avoided due to its commercial implications and assuming that such works are eligible for copyright protection, the most sensible approach would appear to be that adopted in jurisdictions such as the UK where this is attributed to the programmer (as author) by extension. Possibly, attributing ownership to the AI itself may also be somewhat entertained at some point given that citizenship for AI is being considered in Malta, but that would require a long in-depth study of its ramifications and in the meantime, the creative industries need this legal ambiguity addressed yesterday.

Article by Dr Terence Cassar.

For more information on intellectual property laws or the regulation of artificial intelligence in Malta please contact Dr Ian Gauci on or Dr Terence Cassar on

Disclaimer: This article is not intended to impart legal advice and readers are asked to seek verification of statements made before acting on them.