Abstract

Artificial intelligence (AI) systems are creative, independent, autonomous, rational, unpredictable, accurate, evolving with each passing day and the use of it is also increasing in our lives day by day. AI and machine learning has already reached to a point where we hear about independent driverless cars, AI lawyers who do the data analysis of cases and tell the probability of winning a cases, automated dispute resolution systems, automated contracts, social policymakers etc. We have heard about the machines making original work of art, almost all the works that are copyrightable such as paintings, music, poetry, writing news, stories, literature etc. today it is impossible to imagine any of the work which an AI cannot produce. These developments obviously give rise to the issues of authorship, ownership and infringement of copyright for the automated AI. This paper will deal with the intersection of the autonomous AI systems and Intellectual Property Laws, mainly copyright regime.

Authorship is the most fundamental issue of copyright; ownership is also dependent on the authorship of the work. So, authorship of copyrightable works created by autonomous AI is a burning topic in the world. This paper will address the issue and talk about the ways of determining authorship and ownership of copyrighted work generated by AI as providing ownership rights is important for the future of the industry. No country till now has opened their copyright laws to include AI generated work explicitly. Human author is a requirement for all the protection of copyright law till now. But there are two ways of trying to talk about the authorship of AI generated work. First, we can define the term author and include AI into it and secondly we can define the employer- employee relationship between the AI and it’s generating humans or companies and then provide rights accordingly. The work for hire scenario is also explored in this paper where the AI would be a hired system to work or create. The paper address the question of copyrightability i.e. the ownership and authorship of the works generated by the autonomous AI.

1.     Introduction

Artificial Intelligence (AI) has become reality of our generation now, much have been over a decade now is written over the growing capacity of AI in knowledge based fields. Some progress of the systems that we hear every day are Deepblue IBM’s AI system which defeated humans in chess, Hasan Robotics Sophia, Watson etc. By definition AI is a computer system which can perform tasks normally requiring human intelligence. Now these systems which are considered to have AI is owned by real persons or legal entities. The Personhood of the AI machines or agent is under question. Although Saudi Arabia has awarded citizenship to a robot named Sophia, as the ownership over such intelligent systems is open for debate in future with regards to robots rights. We obviously see AI distinct from human intelligence in many area and subjects, one of which is art, creativity, creation of original things. It is argued by many critics to AI that machine lacks creativity as they have bound behaviour as creativity means unpredictability which AI cannot deliver. But AI recently covered by machines trying to mimic human intelligence as the AI system is based on neural networks now which can behave unpredictably such that their original programmer would also not know how they are functioning. These AI systems are acting autonomously to create work of art, giving rise to issues of copyright. Copyright is an integral part of intellectual property rights, it is a legal right provided to a creator of an original work including economic and moral rights.

Through this paper the author wants to highlight legal issues that have come up with the increase in technology in machine learning and AI in Intellectual Property discourse focusing on copyright laws. In paper is divided into two parts the first part describes two problems which has arisen first related to authorship, who can be author and hence can enjoy the economic and moral right of the copyright. And second part deals with copyright ownership of the AI generated work and the beneficiary for it, who will ripe the benefits of the i.e. the holder of economic rights for the computer generated work. It will also include suggestion for providing intellectual rights to the creation of AI systems and conclusion.

2.     Originality of the Copyrighted Work

To determine whether the work is new or original the material should not be copied from the easier existing works, this is evaluated globally even though the protection awarded will be territorial in nature, but the threshold for the same depends upon the countries. There are three major theories of originality

1.     Sweat and brow theory- Which simply means any piece of work can be considered as original on which efforts has been put by the creator or author. This is very low threshold of originality as it will give protection to anything on which labour is done. Mainly simple compilations, tables etc.

The threshold which was created by sweat and brow rule was providing too much protection. Unlike patents where industrial application of the proposed patent application is necessary copyright doesn’t regulate the quality of work and hence due to low threshold many non-creative materials were getting the protection.

2.     Skill Labour and Judgment theory- This means for getting a copyright protection one should have applied his skills and judgment for creation of the material and not mere labour. Mere putting labour cannot ensure protection for the author.

This threshold was used successful for a long period but was creating confusion in protection of databases, directories with the invent of new technologies. It was exact opposite to the sweat and brow theory and was extreme on putting high threshold for originality in copyright protection.

3.     Non-trivial, Non-mechanical Application of Skill, labour and Judgment theory- This means copyright protection can be enjoyed by those material which has created the copyrightable material applying his/her labour, skills and judgement but the skills and judgments applied should not be trivial and should be non- mechanical. Like the mass production of phone numbers arranged alphabetically is a trivial and mechanical application of skill and judgment. We can compare this requirement with that of non-obviousness in patent law for our understanding. The invention should not be obvious to person skilled in the art and in copyright the material should not be so trivial or mechanical reproduction of material i.e. the obvious reproduction of earlier material, but of course the threshold in patent law is very high. This doctrine showed a middle path approach for exercising the skill and judgment in creation of a material for copyright protection.

India has discussed the similar threshold to be followed in Eastern Book Company v. D.B Modakcase, where the question before the court was how much work on a judgment can make it copyrightable?

The courts adopted minimal degree of creativity as threshold for copyright protection where mere labour i.e. copy-editing, clerical corrections, syntax would not suffice to get the protection. A minimum amount of creativity should be involved which is not trivial and non-mechanical in nature, and hence the courts held that headnotes, footnotes and editorial notes of the cases will get copyright protection. The court holds that-

Creative works by definition are original and are protected by copyright, but creativity is not required in order to render a work original. The original work should be the product of an exercise of skill and judgment and it is a workable yet fair standard

3.     Copyright and Artificial Intelligence

As we understood that for getting a copyright, the work has to be original and by the test mentioned above an AI system can easily create an original work of art which is eligible for copyright protection. For understanding the herein mentioned judgments are seen-

i.                 Burrow Gilles Lithorgraphic Co. v. Sarony

The case involved a question that whether a photograph is a mere mechanical reproduction of art or an original work of art in itself and hence a copyright protection can be granted to photograph or not. The court discussed between creative and mechanical labour.

This was a suit for infringement of a copyright in photograph by Oscar Wilde, bought by lithographic company and the court discussed that mechanical labour is not per se creative and hence cannot be protected by copyright. The camera used was discussed as a tool which aided the author in creating the original work and hence the production of a photograph was discussed can be mechanical as argued by the defendant.

The Supreme Court in 1884 also justified the legal standing of providing of a copyright to the photographer. Since than a lot have changed in photography be it digital camera that can work in automatic mode or smart phones. The image is a creation of a AI program which assisted a photographer a human to get the creative work. But by this judgement the court narrowed the possibility of granting copyright to products which are output of machines only as it would be mechanical in nature.

The computer programs responsible for autonomously generating creative work are not copyrightable in any part of the world till now, if it cannot be traced back to a human influencer.

ii.               Bleistein v. Donaldson Lithographing Co.

The facts of the case involve a company engaged in lithographic business alleging that the defendants copied chromolithographs prepared by its employees to advertise a circus. The issue involved was whether or not the lithograph was protected under copyright laws.

This case was in continuation of the previous case. The court in this case very precisely differentiated between a human’s work and something artificial. The court held that the work is entitled for copyright protection. Justice Holmes on behalf of the majority clearly wrote about the uniqueness of the human personality and stipulated the same as a pre requisite for copyright protection, and by extension they meant that there is no scope for anything that is not created by a man’s creativity for protection under copyright law.

iii.             Alfred v. Catalada Fine Arts, Inc.

This judgement among all saw a softer approach to the interpretation of the copyright law. The courts lowered the threshold and standard for originality and held that the work to be original must not be copied from any other work of art and also unintentional or accidental variations may be claimed by the author.

iv.             Naruto v. Slater

In this case a photograph was taken by a Monkey named Naruto and the copyright claim was there, as the animal didn’t had any legal standing in court the court held that an animal cannot pursue copyright, also denying any claim of the copyright to David Slater who owned the camera by which the picture was taken. Which means that the courts denied copyright to any non-human and also put the creative work in public domain.

These four judgments along with the judgement which lay down test for originality clears things a bit ambiguity that prevails around grant of protection of AI systems generated work and the lack of definitive stance will affect the right holder. The shortfall can be lack of incentive in any way for contributing to the development of the AI if no protection is given and everything falls under public domain.

The other approach which is not yet accepted by courts or laws in any of the country is redefining the term author within the copyright law and include non-human entities as well in it. As the right arising out of authorship cannot be solely utilised by non-humans like AI itself hence the authorship can be granted to them and the ownership can remain with creator of those AI whether in a work for hire set up which I will explain later in this paper or by the reason that the creator of the AI system is the developer and owns the AI.