OJAVC Releases Open Framework for AI-Assisted Provenance Research in Cultural Heritage
The Onyx Journal of Arts & Visual Culture (OJAVC) has released the first open, peer-reviewed methodological framework for AI-assisted provenance research in cultural heritage institutions. The protocol, developed by an international working group of curators, computer scientists, and legal scholars, is published as a 48-page position paper accompanied by a fully open-source toolkit.
The framework establishes minimum documentation standards for any attribution claim that relies on machine-learning models, including mandatory disclosure of training data provenance, model uncertainty intervals, and a structured pathway for human expert override. It is designed to be directly compatible with the CIDOC-CRM ontology and the LIDO metadata schema already in use by major museum networks.
Early adopters include the Stedelijk Museum Amsterdam, the Power Institute Sydney, and a consortium of three university museums in Latin America. The full text, model weights, and annotated case studies are available at arts.onyx-press.org under a CC-BY-SA 4.0 licence. A live webinar introducing the framework to museum professionals is scheduled for 5 September 2026; registration is free and open to all.