Talented artists are exceptional, but how do we comprehend great works of art designed by artificial intelligence? Art has been taught to AI systems based on works which can be observed and then used to generate new images. When researchers at Rutgers University tested AI-created artworks against artist-created works, art appreciators were unable to distinguish the artist from the AI-created images between 48 and 85% of the time, depending on the genre. If value is found in works created by Artificial Intelligence, who owns these works and, are they able to be protected by copyright laws? These questions may have seemed crazy twenty years ago, but must be seriously considered today.

AI-Created Works of Art

Scientists have expanded the use of artificial intelligence from simple bots and computer learning to recreate established works of art and music. This has gone even further with the development of a generative adversarial network (GAN), which combined two AI neural networks, one which produces images (called a generator), and one which judges established paintings from a gigantic library of works from all genres, styles, and time periods (discriminator). This allowed the system to “learn” how to create works of art through observation, and develop skills by appraising the work when compared to proven artwork. As already stated above in the Rutgers study, art enthusiasts liked the works, finding them inspiring and complex. In fact, AI-created works have brought thousands of dollars at auctions, proving that they are valuable to the right collectors. However, it is not human-created, as compared to more traditional digital art techniques.

Naruto Copyright Case

In 2008, a British nature photographer visiting Indonesia to photograph endangered Celebes crested macaque monkeys found that his camera had been taken and inadvertently used by one of the monkeys, who proceeded to take hundreds of photographs, including some of himself. After Wikipedia published one of the photographs, indicating clearly that the photo was not under copyright, photographer David Slater began legal proceedings in US Copyright court, stating that he held exclusive rights to the photographs.

PETA stepped in, filing a lawsuit against Slater, demanding that they administer all the proceeds from the photograph use, to be used to protect macaques in an animal preserve in Indonesia. The case was eventually settled out of court, with Slater agreeing to donate 25% of future revenue to protect the monkeys. This settled the question of whether animals can have copyright protection, even if a human owned the equipment, as in this case, a human did not take the photograph.

Shift from Animal to AI—Copyright Protection

Although copyright law doesn’t specifically address artificial intelligence, or even human authorship of original art, it has always been assumed that humans were behind any works. According to Art Law Journal: protection under the Copyright Act must meet the following requirements:

1) an original work of authorship,

2) fixed in a tangible medium,

3) that has a minimal amount of creativity.

If a work of art doesn’t meet all three of these requirements, then it does not qualify for copyright protection. Therefore, it appears that copyright lawmakers have considered the issue in the writing of the law.

Meeting the Copyright Standards

With many AI-generated works of art a combination of existing works and creative modifications, is the work unique enough to qualify as ART? Is it a unique, one-of-a-kind creation? If AI-created works could not be predicted, then they would be considered original. Once the work is “issued” or made public, then it is fixed in a tangible medium.

Copyright protection in art is given as soon as the work is created, so the creator has exclusive right to decide the future use of the work. But a machine-created work can be duplicated easily, so ownership is not clearly distinguished. The copyright law does address computer-generated works:

503.03(a) Works-not originated by a human author.

To be entitled to copyright registration, a work must be the product of human authorship. Works produced by mechanical processes or random selection without any contribution by a human author are not registrable.

Then, Who Owns the Copyright?

Since the latest GAN-created works of art are developed with deep machine learning, where the network creates its own work, one can say that humans did not contribute to the work. But did they? There are a variety of locations within the process of creating art by artificial intelligence where humans took part. Some of these ownership options are:

  • Software builder
  • All the artists whose works the machine studied to create the unique piece
  • Hardware team
  • Company who sells software
  • Owner of the machine which distributed the work

Humans make the decision to make use of the art created by a machine; therefore, a machine clearly cannot hold a copyright since it cannot give permission for the use of the work. The monkey lawsuit was deemed frivolous for this reason. If a person or team contributed enough inputs which could be considered creative enough in the formation of the work, then they would own the copyright. If the design, however, is attributed mainly to the program (artificial intelligence), then the work is not copyrighted and belongs purely in the public domain.

The future in terms of legal protection of these works is still questionable, as we truly do not know the full extent to which AI will be used in creative works in the future.