Future copyright problems

Artificial intelligence (AI), the buzz word of 2017, is the simulation of human intelligence processes by computer systems and other machines. AI has advanced rapidly over the last few years, with applications ranging from the novel and benign; like morphing children’s drawings of their parents into pictures of real people; to the complex and useful, such as autonomous vehicles and Apple’s new Face ID system. As tends to happen, however, the law may be playing a game of catch-up with the advancements in AI.

Audi, for example, recently released their new flagship luxury model – the new A8. The A8 is the first vehicle to offer level 3 autonomous driving. This means that the vehicle can pretty much navigate traffic by itself while the driver streams a movie, or gazes out the window at the proletariat. When the A8 goes on sale, however, it will not be sold with all of the level 3 autonomous driving hardware installed. The reason being is that China, the USA and a host of other countries are still negotiating the legal frameworks around level 3 autonomous driving.

The law of copyright may also currently be one step behind the advances in AI. Broadly speaking copyright, governed by the Copyright Act, protects certain defined “works”, such as computer programs, films, literary, musical and artistic works. Traditionally, the creation of these works has been a strictly human endeavour. The advances in AI, however, have resulted in a paradigm shift, in that sufficiently “smart” computers can now create their own “works” with little, or even no, human involvement.

For the thriller film Morgan (debuting 2 September 2017 in the USA), the IBM Watson supercomputer was tasked to make the scariest movie trailer possible. In making the trailer, Watson analysed the visual and audio composition of 100 classic horror movies, to learn what is “scary”. Using this new knowledge Watson extracted scenes, dialogue and music from the entirety of Morgan to create what, in its “opinion”, is the scariest movie trailer possible. This trailer was created by Watson using multiple sources of data, ostensibly input by multiple humans as well as new information Watson itself created.

Requiring even less human involvement, the artist Taryn Southern composed her new single “Break Free” using an open source AI platform called Amper Music. Personally, this sounds like something out of George Orwell’s ‘1984’, as Southern merely selected the genre, the instruments and the beats per minute, Amper then created the musical verses. The musical verses were then arranged by Southern into the backing track, which were then layered beneath her vocals.

The above movie trailer and pop song were created by AI, or at least by using the input of AI which brings us to the importance of the distinction between a computer aided work and a computer generated work. The former, is work made by a person using the computer as a tool, or instrument. The latter, is where a work is made by a computer where it is not possible to attribute the resultant work directly to the efforts of any person causing the work to be made.

The authorship and ownership of computer aided works raise far fewer issues than computer generated works so, for current purposes, computer generated works will be the focus. Computer generated works, raise some interesting questions in terms of copyright to which there are currently no clear answers, these are:

Who would be the author of a work created by an AI?

Section 1, paragraph (h) of the Copyright Act, under the definition of “author”, states that “if a literary, dramatic, musical or artistic work or computer program is computer-generated” the author would be the person “by whom the arrangements necessary for the creation of the work were undertaken”. This phrase is, nevertheless, somewhat problematic. Is the person “by whom the arrangements necessary for the creation of the work were undertaken” the person who built the AI, the person who trained the AI, or the person who fed the AI specific inputs to the AI?

Who would be the owner of a work created by an AI?

In many instances, once the work has come in to existence the authorship and ownership of copyright subsisting in it coincide. This, however, is not always the case with the exceptions being set out in section 21 of the Copyright Act. That being said, as authorship of a work created by an AI is imprecise, it follows that ownership will also be difficult to determine.

Who would be responsible if an AI creates a work that infringes on another’s copyright?

There is currently no legal framework that provides for an AI as an infringing party or even if an AI can be an infringing party. Therefore, should an AI create a work which infringes on another’s copyright, it is unclear who would be held accountable for such infringement. It may be the person “by whom the arrangements necessary for the creation of the work were undertaken” but, amongst other issues, this “person” is very hard to accurately ascertain.

As things currently stand, we are in uncharted waters in regulating and managing AI created works, without the certainty required to address these new legal concerns. This leaves policy makers and industry partners with some serious and complex work to develop a legal framework that is both flexible and comprehensive.

Perhaps we can request Watson to do it for us.

This article was first published in Business Day.

Date published: 9 October 2017
Author: Brendon Ambrose

Tags: copyright artificial intelligence