Legal framework for the civil circulation of ai-generated content and prompts: a contracts-based approach

Abstract

The article provides a comprehensive civil law analysis of contractual models for regulating the relationships that arise in the course of civil transactions involving the results of the operation of artificial intelligence (AI) systems and control instructions (prompts). The article examines the problem of an institutional vacuum resulting from the absence, both in Ukraine and in other countries, of a public-law mechanism for the registration of non-original objects generated by computer programs and for the establishment of sui generis rights. It is argued that, in the absence of a state title and a public presumption of rights vis-à-vis third parties (erga omnes), a civil law contract transforms from an instrument for the transfer of rights into a mechanism for establishing rights and quasi-registration, which operates within the framework of an obligation (inter partes). The objects are differentiated into the categories of AI-assisted (created using AI with human creative input) and AI-generated (autonomously generated by algorithms). It has been demonstrated that the copyright paradigm applies to AI-assisted works, incorporating representations and warranties regarding the auxiliary nature of AI and indemnity mechanisms. In contrast, for AI-generated works, a shift in the contractual paradigm is highlighted: a transition from contracts for the transfer of rights to service models (Prompt-as-a-Service), the transfer of actual control, and confidentiality agreements, which are important in cross-border transactions. Standard quasi-registration terms for contracts are proposed, which provide for the detailed recording in specifications of timestamps, prompts, model parameters, and hash codes, thereby replacing the registrar’s stamp. The paper justifies the protection of prompts through the concept of trade secrets and the use of adapted “work-for-hire” clauses with a condition of immediate assignment of rights to prevent the leakage of generation tools