One of the main contentious points when artificial intelligence, deep learning or machine learning (for the distinction between these functionalities, see here) are used for generating creative works is the question of attribution of works to an author, usually a human who has used the tech tool with some level of involvement in the creation (for the debate around the protection of AI generated works, see here).
In an interesting case (Court of Cassation, Civil Section 1, order 1107/2023), the Italian Supreme Court has left the door open for the protection of a graphic work generated by a machine learning tool, at least where the parties have never discussed in the merits whether and to what extent the use of such tool had absorbed the creative elaboration of the artist who had made use of it. This is a two-part post, whereby this part looks at the Italian case, while Part 2 will focus on the same issues but from a US perspective.
The Italian ‘floral fractal’ case
This case concerned the commercial exploitation of a graphic work depicting a floral composition, adopted as a fixed background during the 2016 edition of the Sanremo Festival. Specifically, the work in question is a floral fractal created with the Apophysis software, which has been called “The Scent of the Night“. In general, a fractal is a figure characterized by “self-similarity”, or by the repetition of its shapes on different scales of magnitude. Basically, the software generates the fractal, which is then used by the author to create the work – in this case, a personalized flower. Apophysis is a tech tool, implementing machine learning techniques, with many features for creating and editing fractal flames, including an editor which allows one to directly edit the transformations by manipulating triangles, a mutations window, which applies random edits to the triangles, and an adjust window, which allows the adjustment of colouring and location of the image (here and here).
The judgments in first and second instances
On 18 July 2018, an architect summoned the Italian public broadcaster RAI, claiming that the copyright in her graphic work “The Scent of the Night” had been infringed. She requested compensation for damages, as well as the removal of the program from the RAI website and the publication of the judgment.
In the first instance, the Court of Genoa with a judgment of 6 June 2018 attributed the authorship of the work to the claimant and confirmed violation of the copyright by RAI, which has made commercial use of the work in the context of the well-know Sanremo Festival. According to the Court,the ownership of the work could be attributed to the claimant on the basis of the prints of the websites she produced and via an online search, as well as a book published by Mondadori which contained the image with the attribution to the architect. In addition, the Court held that the work was to be considered creative.
On 11 November 2020, the Court of Appeal of Genoa confirmed these conclusions, stating that creativity cannot be excluded just because the work consists of simple ideas and notions, included in the intellectual knowledge of people having experience in the subject. Moreover, according to the Court, creativity is not constituted by the idea itself, but by the form of its expression, or by its subjectivity, so the same idea can be the basis of different works, which are or can be different due to the subjective creativity of each individual author and which, as such, is relevant for the purposes of protection. In this case, the Court of Appeal observed that the work is creative when it expresses an original elaboration deriving only from the inspiration of its author and confirmed the assessment of the first instance judge, arguing that the image was not a simple reproduction of a flower, but involved a genuine reworking of it, and was therefore worthy of authorial protection due to its creative nature.
In addition, the Court of Appeal strengthened this reasoning by referring to the broad appreciation given to the graphic work by RAI itself. This is reflected in the presentation of the event to the press, which highlighted the fractal flower and its symbolic value by making it an important choreographic component on stage, instead of sticking to the traditional decorations of real flowers. Finally, the Court of Appeal considered the degree of notoriety achieved by the work on the web as a further confirmatory indication.
The Court of Cassation
Before the Supreme Court, RAI argued that the first and second instance courts erroneously qualified as a creative elaboration a software generated image which lacks any creative degree attributable to its supposed author. RAI alleged that the work of the architect is a digital image, with a floral subject, with a so-called ‘fractal’ figure, i.e. characterized by “self-similarity”, or by the repetition of its forms on different scales of size and was processed by software which elaborated its shape, colours and details through mathematical algorithms; the alleged author, i.e. the architect, only chose an algorithm to apply and subsequently approved the computer-generated result.
The Court of Cassation considered this argument “new” and inadmissible because it was not introduced in the first or the second instance. To this end, the counterparty’s admission of having used software to generate the image is not sufficient for allowing the Supreme Court (as judge of legitimacy and not of merits) to further investigate this argument, since such admission is still compatible with the elaboration of an intellectual work with a sufficient degree of creativity. A factual assessment would have been necessary to verify whether and to what extent the use of the tech tool had absorbed the creative elaboration of the artist who had made use of it.
The Supreme Court’s judgment is notable for two reasons: (i) it has not excluded the possibility of copyright protection for machine learning generated works with an adequate level of involvement of a human author and with a factual scrutiny to be conducted on the merits, and (ii) it has considered the case as part of the wider topic of so-called digital art (also called ‘computer art’) to be further explored in its jurisprudence.
The Italian Supreme Court in RAI vs. Biancheri has stated that the mere argument of being a graphic work generated via machine learning does not exclude per se the protectability under copyright laws.
Nonetheless, the judges have expressed that there is still a lack of understanding in the rules applicable to AI-generated works and have clarified that there is the need for a factual analysis to evaluate the level of human intervention in the creation of graphic works via AI systems, in order to allow access to copyright protection for such works.