The AI Discussions Working Group is a GLAM community initiative that shares information and resources and provides a neutral space to discuss the evolving impact of AI-driven bots on institutional infrastructure.
Our Purpose
The AI Discussions Working Group seeks to provide support to academic, cultural heritage, and collecting institutions experiencing disruptive system impacts from large-scale AI bot activity, including scrapping of digital collections for large language model (LLM) training.
Participants are typically, but not solely, represented by individuals responsible for maintaining access to these systems. Our group includes representation from repository managers, developers, systems administrators, academic library leadership, and a variety of other technologists, many of whom are often responsible for responding to these challenges without sufficient institutional resources, guidance, or visibility.
With a focus on many of the technical mechanics of AI scraping and mitigation strategies, the group also prioritizes community building, shared learning, and advocacy. We aim to create a trusted, collaborative space where participants can exchange experiences, identify emerging patterns, and collectively develop resources that help institutions understand, mitigate, and respond to these impacts.
Resources
How to Get Involved
We invite all interested individuals to join the AI Discussions Working Group. All skill levels and experiences are welcome to engage in our monthly discussions, participate in the Solutions Showcase Series and contribute to our shared resources.
Monthly Meetings
Currently, the AI Discussions WG meets monthly - on the 1st Monday of each month at 11am Eastern.
- Note: In order to respect participant confidentiality, and encourage candid discussion among attendees, meetings are not recorded and no formal meeting minutes are taken.
If you would like to be added to the recurring meeting invitation or stay informed about on-going activities, please request to be added to the Communications List by using the following form: https://forms.gle/dkUuyZVVrjF9FVZq8
Solutions Showcase
Want to see a Past Recording?
Recordings of past Solutions Showcases are available upon request. They may be shared with colleagues and peers, but please do not post them in publicly accessible spaces, as this could expose the mitigation strategies to the bots they are designed to prevent. 
To get a copy of a recording, please submit the following form: https://forms.gle/zwkujzxZh2pBNCgh7
Have a solution to share?
We’re actively looking for additional presenters! If your institution has addressed AI bot scraping in a creative or effective way, we’d love to feature your work. Please reach out to Arran Griffith at arran.griffith@lyrasis.org to book a presentation slot.
Background
The AI Discussions Group emerged in response to conversations early in 2025 between members of Fedora Governance who were experiencing a new wave of service and system disruptions within repository systems. Behaviors included issues like sustained spikes in automatic traffic from foreign and distributed IP ranges which were impacting repository performance and availability.The activity was inconsistent, difficult to predict, and increasingly disruptive to both systems and staff workflows.
As these concerns were shared more broadly, it became clear that Fedora was not alone. Similar experiences were being reported by technologists working with other repository platforms and across a wide range of academic and cultural heritage institutions. Many individuals were addressing these challenges in isolation, without a shared vocabulary for describing the problem or a clear path for escalating concerns within their organizations.
In response, Fedora community leadership convened a series of open discussions with members of the Samvera and Islandora communities to determine whether others were encountering similar impacts. Interest in this conversation was immediate and significant. Participants from across the GLAM ecosystem, from a variety of institution sizes and types, joined to share experiences, compare observations, and seek mutual support.
These early conversations quickly evolved into a recurring, open forum characterized by confidentiality, trust, and peer-driven dialogue. Participation grew quickly, and common themes emerged. The group recognized the need to move beyond discussion and understand how to best serve our individual communities through actionable output. The AI Discussions Working Group was informally established to provide structure to this work, support collaborative resource development, and advocate for the technologists and institutions navigating the evolving impacts of AI-driven harvesting of digital collections.
