Workshops highlight standards needed for tracking provenance, attribution, and usage in the AI ecosystem

In May 2026, NISO, COUNTER, and Cambridge University Press & Assessment hosted a series of invitation-only workshop discussions exploring how the information standards community should adapt to the growth in use of agentic AI technologies for discovery and delivery of content. 

Traditionally, the value and impact of scholarly content have been measured by the number of human downloads and views. However, as AI agents ingest content to generate search results, human readers can now access information without ever landing on a publisher site. AI agents also operate at the component level, whereas current systems support metrics at the article, title, or chapter level. This shift from human to computer readership has resulted in new challenges in tracking the provenance, attribution, and usage of scholarly content.

During the discussions, a consensus emerged around several ways in which the community should adapt. A first step in addressing the questions of provenance and attribution is developing a minimum provenance payload (MPP), or metadata fields for scholarly content (including at the component level) supplied to an AI system. New standards are also needed for managing AI access to content. Finally, to better understand usage patterns, COUNTER will continue to develop best practice guidelines for tracking usage at the sub-item level and engage technology companies in developing a shared reporting structure.

Read the synthesis report, “Standards for Provenance, Attribution, and Usage in the AI Ecosystem,” for additional details and next steps.