Copyright and Artificial Intelligence: New Challenges

by Jaqueline Simas de Oliveira

December 01, 2017


One year has passed since a Dutch group made public the painting entitled “The Next Rembrandt”. They uploaded several images of paintings of the famous 17st century painter, Rembrandt, in an Artificial Intelligence (“AI”) software.  After processing and comparing these images, the software autonomously created a new painting, which, under the view of any critic of arts, could have been created by Rembrandt himself.

This case is a perfect example of the digital revolution we are experiencing. This revolution tests the boundaries of a number of consolidated concepts and poses new challenges for all areas of human knowledge, including copyright law.

Given this scenario, the creation of works of art through AI is substantially increasing and raises an interesting question: who is the owner of the copyright over AI-generated Works?

The use of computers for creating copyrighted works is not something new. For years, artists have been using software to sketch, retouch and finalize their works. Books are written in Word, architectural projects are created with AutoCAD and photos are edited in Photoshop.

In these cases, the role played by technology is clear: a mere tool that helps the artist to fix its work in a tangible medium. It is also clear that the rights related to these new works of art are owned by the user of the software, i.e., by the one who manipulated the functionalities provided by the program.

On the other hand, when it comes to AI, the creation of artistic works can go far beyond the mere fixation of the work in a tangible medium. In the cases wherein the artist makes use of the AI as a mere tool, it is clear that the consolidated understanding related to the use of ordinary software would apply as well.

An important discussion inaugurated by AI, however, is related to cases where the conception and fixation of the work occur regardless of human intervention.

Under Brazilian law, Article 11 of the Copyright Act provides that the author is the natural person who created the work. This means that our current legislation does not allow copyright to be held by a machine.

In the same sense, the United States , Australia  and the European Union  have, in multiple occasions, adopted the understanding that the protection conferred by copyright can only be granted to works of art created by human beings.

This scenario raises the argument that works autonomously created by artificial intelligence and without human intervention, would already be born in the public domain.

Nonetheless, this proposition seems inappropriate, as far as it could represent a discouragement to the development of new technologies related to IA, especially in the artistic field.

After all, what would be the interest of companies to invest in the mentioned area if, at the end, they do not own any rights over the final result generated by the AI program?

Thus, the granting of protection to this new artistic works may be relevant, particularly to some sectors of the economy. Real examples of this are startups  of the entertainment industry which are already using artificial intelligence to create music to be used the field of video games.

Considering the above, it seems that countries engaged in the protection of intellectual property rights cannot ignore this new reality and fail to regulate the matter. However, what would be the most appropriate way to regulate the copyright related to these creations?

Although Brazilian law does not yet regulate this theme and practical discussions about the subject are still incipient, the foreign experience has some interesting examples of the possible treatment to be given to this matter.

The UK Copyright, Designs and Patents Act (CDPA) provides, for instance, in its section 9 (3) that "in cases of creation of literary, dramatic, musical or artistic works by computers, the author shall be the person who made the necessary arrangements for the creation of the work in question.”

In addition, section 178 of the CDPA states that “works created by computers, are those created in circumstances where there is no human author."

As seen above, the CDPA assigns the copyright of artistic works created by AI to the person who made the creation of the work by the computer feasible. In the case cited at the beginning of this article, it seems clear to us that under this system, the copyright of the work would be granted to the group that enabled the machine to create "The Next Rembrandt".

However, this system still yields discussions, as far as it will only possible to determine who is the person responsible for making the creation of the artistic work feasible on a case-by-case basis.

After all, it is not necessarily the end-user of the AI that enables the creation of the work, whose responsibility in some cases may be limited to activating the machine. In these cases, the author of the work may be the one who programmed and fed the system with the content that, once processed, made possible the creation of the artistic work.

Given the above scenario, it is clear that the Brazilian legislation need to be adapted in order to guarantee the protection of works conceived by AI. This could be done either by modifying the current Copyright Act or by creating a sui generis right, treated by a specific statute.

In this sense, Congress should be careful to enact rules that clearly identify the owner of the AI-generated work, in order to mitigate litigation and provide legal certainty in an area that will be extremely relevant in the upcoming decades.


Jaqueline Simas de Oliveira


read +

related posts