Abstract:
Contracts have been proposed to complement copyright law to better regulate the use of data in training AI models. Still, the dynamic between contracts and copyright law in this context remains underexplored. This article examines their complementary and conflicting dynamics, identifying an “emerging contract override” across various scenarios, which includes how contracts interact specifically with copyright exceptions, how contracts serve as a mechanism to create rights, their role in the opt out and opt in debate, and their interaction with the transparency requirement. The analysis shows that regulating AI data through contracts is fragmented, temporary, and uncertain – like building a sandcastle at high tide. While contracts enable tailored private ordering, development in copyright law, competition law, and technological advances can easily erode their enforceability and coherence.