The National Information Standards Organization (NISO) has launched a 12-month pilot project to develop a minimum viable metadata model for tracking the provenance and attribution of content used by artificial intelligence (AI) systems. The initiative aims to address challenges in identifying the sources of information retrieved and processed by AI tools in scholarly publishing.
As AI-assisted search becomes more widespread, AI agents increasingly retrieve and analyze large volumes of published content and data to respond to user queries. This shift has contributed to a zero-click environment in which users obtain information from published sources without visiting publishers' websites. The trend creates challenges for scholarly communications, as publishers and librarians have traditionally relied on usage statistics to evaluate content engagement and value. Additionally, large language models can struggle to accurately identify and cite the sources of the information they provide, raising concerns about transparency, credibility, and trust in scholarly research.
The NISO pilot will develop and test a minimum metadata model designed to capture provenance and attribution information when AI systems retrieve content or data from scholarly sources. The project is intended to establish a foundation for more consistent identification of content sources within AI-driven information environments.
The initiative builds on stakeholder discussions about provenance, usage, and attribution in AI systems that began in 2025. In February 2026, the NISO Plus Baltimore meeting featured two preconference sessions focused on these issues. In spring 2026, NISO, Cambridge University Press, and COUNTER co-hosted workshops to examine potential solutions. A report summarizing the workshop discussions and identifying standards needed for the AI ecosystem was published in June 2026.
Following these discussions, NISO assembled a group of content creators, publishers, repositories, AI tool developers, and end users to develop a minimum metadata payload for communicating provenance and attribution information. The work is also intended to support further standards development in this area.
The project is funded by participants in the Trusted Retrieval and Attribution for Content Ecosystems (TRACE) initiative, a cross-industry collaboration focused on developing infrastructure to support the use of AI systems in research publishing. The metadata model is one of several TRACE workstreams. Other activities include tracking AI-related content usage through COUNTER and studying how AI systems interface with scholarly content.
Todd Carpenter, Todd Toler of Ithaka S+R, Tasha Mellins-Cohen of COUNTER, and Priya Madina of Springer Nature introduced TRACE at the STM meeting in Frankfurt on October 5, 2026, ahead of the Frankfurt Book Fair.
Further information about TRACE and its associated projects, along with NISO's metadata pilot, is available through The Scholarly Kitchen, which published a related post by Todd Carpenter on October 6, 2026.
Click here to read the original press release.
