Managing Access to Open Repositories in the Age of Generative AI

The rapid rise of generative AI and its reliance on extensive training data has led to heightened interest in open access repositories, which house legally and freely available as well as trustworthy open knowledge. While training Large Language Models (LLMs) aligns with Open Access and Open Science principles and while it provides a promising avenue to mitigate AI’s well-documented shortcomings, such as the potential for spreading of misinformation, they have also exposed open repositories to surging web crawler traffic. This phenomenon poses significant challenges, including potential system overload and resource strain, leading to a heated debate within the community on how best to manage machine access. This panel will explore the pressing questions surrounding the intersection of open access principles and generative AI demands. Should repositories implement stricter controls on web crawlers? How can repository managers distinguish between ethical and abusive agents, and what technical solutions are available to achieve this balance? Experts representing repositories, indexing services, and interoperability and standardisation efforts will provide diverse perspectives on these dilemmas and will involve the audience in this discussion. By fostering a dialogue among stakeholders, this session aims to identify fair, sustainable, and community-driven strategies that preserve accessibility while protecting repository infrastructure. Participants will engage with thought leaders, contribute to best practice discussions, and help shape a collaborative future for open repositories and generative AI.