GLAMorous AI TL;DR — May 2026

Before the Model: Getting Collections AI-Ready

Author

Alphaeus Lien-Talks
May 26, 2026

"AI can support archival work, but it is not a magic solution." 

Colavizza & Jaillant, FLAME (2026)

Welcome to this months edition, Before the Model, a roundup of how the GLAM sector is shifting its attention from what AI can do for us to what we need to do first. The story across spring 2026 is consistent: preparedness, provenance, and governance now come before any model. Whether the deliverable is a chatbot, a metadata enrichment pipeline, or a biographical note, the limiting factor is rarely the model. It is the readiness of the collection, the documentation around it, and the institutional capacity to evaluate what the model produces.

This month's theme is readiness: the unglamorous, infrastructural work that decides whether AI helps or harms.

Colavizza & Jaillant – AI Preparedness Guidelines for Archivists

A concise, open-access framework built around four pillars: completeness and excluded data, metadata and access, sensitivity and rights, and governance and skills. The message is unambiguous: automation is a constrained necessity, not a magic solution, and preparation is the precondition for responsible use.

🏛️ Readiness as Practice

Fritz – Finding Common Ground in GLAM: Ethical AI in Three Themes 

The introduction to a special focus issue on ethical AI in digital stewardship. Identifies three converging themes across the GLAM sector: an ethics of care, capacity-building, and institutional readiness, framing ethical AI as a stewardship problem rather than a technical one.

Verkerk and Harmon – Navigating the Limitations of AI in Archival Description: Can and Should AI Write Biographical Notes? 

A grounded look at where LLMs help and where they quietly fail in descriptive work. Biographical notes turn out to be a particularly revealing test case: plausible prose, confidently wrong context, and a steady need for archivist verification.

🔗 Provenance and Trust

Library of Congress (Brador) – C2PA for G+LAM Community of Practice (user stories) 

The first major output of the C2PA for G+LAM Community of Practice. Argues that content authenticity and provenance (CAP), long-standing archival principles, are now under pressure from AI-mediated workflows. Calls for proactive adoption of provenance standards so digital collections remain verifiable from creation through access.

🌍 Beyond Description

Jaillant, Mitchell, Ewoh-Opu & Hidalgo Urbaneja – How can we improve the diversity of archival collections with AI? Opportunities, risks, and solutions 

Examines whether AI can help surface under-represented voices in archives, while warning that the same tools can entrench existing silences if collections, metadata, and training data are not deliberately prepared.

Houston – Bringing Hidden Histories to Light: AI, Archives, and the Future of Digital Stewardship 

A practitioner's reflection on testing JSTOR Seeklight against real backlogs. Useful as a counterweight to the white-paper view: it shows what "AI-ready" looks like at the desk, not in the policy document.

⚖️ Governance and the Wider Frame

ARA / FLAME project – New AI Guidelines Launched to Help GLAM Prepare for the Future 

Sector-level framing of the FLAME guidelines, including their role in moving institutions away from "dark" archives towards more accessible, sustainable digital practices, with governance and documentation as the load-bearing elements.

❓ Big Question

If AI readiness is really about preparation, documentation, and governance, then the question is not which model to adopt but whose readiness counts.

Which collections, communities, and institutions get prepared for AI first, and who decides what "ready" looks like?

💬 About

I'm Alfie, a researcher and archaeologist exploring where heritage, ethics, and AI meet. This digest keeps things short, critical, and useful, no jargon, no hype.

👉 Read or subscribe at glamorousai.beehiiv.com 

👉 Send papers, reports, or ideas for June