How does AI affect photography? The legal perspective.

From the invention of photography and the daguerreotype (1839) to the current era of AI, each technology poses new legal challenges, particularly from the point of view of copyright, which requires originality as a condition of protection. Since its invention, photography has been regarded by some as a simple mechanical means of reproduction, while others saw it as a genuine means of expression for the artist. These issues are coming to the fore again with the advent of AI, where images are generated automatically. North American artist Jason Allen, the disputed winner of an art competition in Colorado, spent more than 80 hours finding the right instructions to generate images of Victorian-style costumes with astronaut suits, with the desired lighting and colours, to create his Théâtre D’opéra Spatial (2022). Out of the 900 images generated by the AI, it is said to have selected only 3, the definition of which was improved with Gigapixel AI before being printed on canvas.

The phenomenon raises fascinating copyright issues. Firstly, from the point of view of the output, some artists are claiming rights to these AI-generated creations, while others are refusing such protection on the grounds that there is no creative contribution by the artist. Then, from the point of view of input images, other artists and image banks, such as Getty Images, are complaining of copyright infringement simply because their works are used to feed the algorithm.

Professor Yaniv Benhamou, a lawyer and associate professor of digital law at the University of Geneva's Faculty of Law, specialises in data protection and governance, and creative and technological law. He invites us to explore the history of photography from its earliest days to see how the notion of originality has been transformed as techniques have evolved. The momentum is particularly well chosen. Firstly, because a large number of legal actions initiated in 2023 will make it possible to determine to what extent it is lawful to re-use billions of images to train algorithms. Secondly, because a number of recent rulings redefine the notion of originality of creations generated by AI, distinguishing, for example, between the artist's creative contribution and the degree of control exercised over his or her tools.

Before opening the floor to questions from the audience, Paul-Louis Roubert will be in conversation with Yaniv Benhamou. Paul-Louis Roubert is a historian of photography, HDR lecturer at Paris 8 University, President of the Société française de Photographie and co-founder of the journal Photographica.

This talk is part of the research project Zone grise de l'original, directed by Pierre Leguillon at HEAD Geneva.

Image: Variation of Jason Allen, Théâtre D’opéra Spatial, 2022, realised with dreamstudio.ai