Everywhere in the US, members of the entertainment industry have been on strike for a few weeks, and Artificial Intelligence (AI) is at the heart of their concerns. Indeed, now that AI is able to write scripts and to create highly realistic videos, actors and TV producers are afraid of being replaced.
But legally, can an AI actually create intellectual works worthy of copyright protection? At the moment, both US and Italian legal systems seem reluctant, considering that an author is necessary a human being (A), and that copyright protection may only be granted to works created with sufficient human contribution (B).
A-) As a non-human person, an AI cannot be an author within the meaning of Copyright Law
Before anything else, it is worth considering whether an AI can be legally considered as an “author”. Pursuant to Article 1 of Italian Copyright Law (22 April, 1941, n.633), protection is granted to “intellectual works of creative nature that belong to literature, music, visual arts, architecture, theater, cinematography, regardless of their form of expression”. In the absence of any textual definition, the Italian case law clarified that the wording “intellectual work” does not refer to the idea behind the work (an idea being usually protected by a patent and not a copyright), but rather to the form of its expression, which must be creative. As well established under Italian case law, the creative nature, “does not coincide with that of […] absolute originality and novelty, but conversely refers to the personal and individual expression of an objectivity belonging to the categories listed in Article 1” (for a recent example: Italian Court of Cassation, January 16, 2023, n. 1107). Therefore, it can be “simple ideas and notions“, as long as “the work reflects the personality of its author, manifesting his free and creative choices”. Such an interpretation is in line with European Union law, according to which an original work exists “if it is the author’s own intellectual creation reflecting his personality” (17th recital in the Preamble of Directives 93/98 and 2006/115). Yet, the concept of personality seems to refer only to human persons. In this regard, the European Parliament, in its resolution of 20 October 2020 on intellectual property rights for the development of artificial intelligence technologies, highlighted that by addressing the author’s personality, the very concept of intellectual creation is linked to a natural person, which basically excludes AI technology.
Similarly, US Copyright Act protects “original works of authorship fixed in any tangible medium of expression” (Section 102, a). Here again, as recalled by the US Copyright Office (hereinafter “USCO”) in its Copyright Registration Guidance, it is well-established that copyright can only protect material that is the product of human creativity. Indeed, the Supreme Court often refers to the author as a “person” or a “man” (see for instance §56 and 58 of the case Burrow-Giles v. Sarony, 1884). As a consequence, a monkey cannot register a copyright (see the high-profile case Naruto). If even a monkey lacks humanity within the meaning of the US Copyright Act, one can imagine that the same applies to robots and AI.
However, this does not mean that technological tools cannot be part of the creative process. Photography is the most obvious example: even though he used a camera to take the picture and/or Photoshop to edit it, the photograph remains the author of such a work, as long as it represents his “original intellectual conceptions” (Burrow-Gils v. Sarony, §58). The same applies in the EU framework: photos can be afforded copyright protection if they are original i.e. if they are the author’s intellectual creation (see Article 6 of Directives 93/98 and 2006/116). Therefore, as highlighted by the USCO, what matters is “the extent to which the human had creative control over the work’s expression”.
B-) The work must contain sufficient human contribution to be granted copyright protection
The European Parliament insists on “the difference between AI-assisted human creations [which might be copyrightable] and AI-generated creations”, which are, at the moment, not copyrightable.
To determine whether an AI-assisted human creation is copyrightable, the USCO recommends a case-by-case inquiry. It must be assessed whether the “traditional elements of authorship in the work” are conceived and executed by a machine or by a human. If an AI generates material without or with no sufficient human involvement, the Office will simply not register it. On the contrary, if a work containing AI-generated material with sufficient human authorship, it is copyrightable. However, in accordance with Section 103(b) of the Copyright Act and with the recent decision Zarya of the Dawn (21 February 2023), copyright will only protect the human-authored aspects of such work, and not the AI-generated material itself. The Zarya case involved a graphic novel combining human-authored text with images generated by the AI service Midjourney. The USCO granted only limited copyright to the novel, considering that the work was indeed protected but not the individual images themselves.
Similarly, the Italian Court of Cassation, in the recent case involving the representation of a flower created thanks to a software (16 January 2023, n.1107, 1st Civil Chamber), recommended a strict scrutiny of the degree of creativity when a software is involved. If so, judges must “assess whether and to what extent the use of the instrument absorbed the artist’s creativity”. If the judge finds that the human contribution is greater than the technology contribution, there is no reason to refuse to grant copyright protection to the person who used such a tool. As opposed to the US, the Italian case-law seems to grant protection to the work as a whole, without distinguishing the parts attributable to the author and the parts created by the AI.
Let’s take a concrete the example: based on the data it was trained on (which may include books, web pages, etc.), ChatGPT produces a complex work in order to answer the prompt entered by the user. Is the output, i.e. the material generated from a prompt, copyrightable? Even though the prompt comes from a human, the USCO considers that the traditional elements of authorship are determined and executed by the technology (in this case ChatGPT) and not by the human user. Therefore, it is not protected by copyright.
Anyway, copyright has been designed to protect the author’s economic rights (allowing him to earn incomes for the use of his work), as well as moral rights (which protect the integrity of the work and the reputation of its author, including for instance the right to remain anonymous). As a non-human being, one may think that an AI does not actually need such a protection, since it will not face economic or moral issues. However, this solution could have a harmful effect on creativity, innovation and development of new technologies: who would be willing to invest in AI technologies if the work they generate cannot be protected? But if protected, who would be the copyright owner? Many questions remain open…
For more information do not hesitate to contact us at the following e-mail address: [email protected].