OPINION: Ten Priorities for Scholarly Publishing through 2027
Darrell Gunter sets out his wishes for publishers, societies, libraries, researchers, funders, and technology partners
I have attended the Frankfurt Book Fair since 1996, and I always look forward to seeing colleagues at the STM Frankfurt Dinner and Conference as well as the Book Fair. The chance to meet in person, exchange ideas between sessions, and continue a conversation over dinner matters to me. I need to be home this year, so I will miss those encounters. These are the topics I would have wanted to discuss with you.
Our mission at Gunter Media Group is to advance humanity by advancing quality research. That ambition asks more of scholarly publishing than making more content available. We must help people find research, judge its reliability, use it responsibly, and understand how knowledge changes when new evidence emerges. As we prepare for 2027, I see ten priorities that publishers, societies, libraries, researchers, funders, and technology partners can pursue together.
1. Make responsible AI a shared practice
AI is already involved in writing, reviewing, editing, discovery, and research analysis. Each organization needs clear governance for acceptable uses, human responsibility, and disclosure. Authors should know what to report; editors should know what to assess; readers should be able to tell where AI played a material role. STM, COPE, the International Science Council, and the Global Young Academy are participating in work toward a common reporting standard for AI disclosure in research. That effort gives us a practical starting point for more consistent expectations.
2. Treat research integrity as a continuing obligation
Paper mills, manipulated images, fabricated references, and compromised peer review threaten confidence in the scholarly record. Screening submissions is necessary, but the obligation continues after publication. Publishers and institutions need clear ways to investigate concerns, correct the record, and communicate changes to the services through which people encounter research. STM’s Integrity Hub illustrates a collaborative approach to detecting problems at submission; BioCore’s Trust Seal initiative offers a complementary effort to recognize trustworthy publishing practices. Neither removes the need for sustained editorial judgment.
3. Clarify rights before AI use becomes routine
Publishers, authors, and technology companies need explicit terms for AI training, retrieval, summarisation, and generated answers. Permission to read an article does not, by itself, settle every question about reuse. Agreements should address attribution, permitted uses, reporting, compensation, and the rights authors retain. Copyright Clearance Center and Creative Licensing International offer different forms of licensing expertise; their work deserves attention as publishers evaluate new arrangements. The goal is a market in which trusted content can be used responsibly and its creators can share in the value it produces.
4. Turn access into usable, verifiable knowledge
An article that can be downloaded has value. Its value grows when a researcher can locate a specific finding, examine the evidence behind it, and follow an answer back to its source. Publishers should test whether their content is structured well enough for precise discovery and responsible AI use. They should require vendors to preserve context, citations, provenance, and permissions as content moves into new formats. The useful question for a pilot is straightforward: Can a researcher verify the answer more easily than before?
5. Make metadata part of the trust infrastructure
A trustworthy research object should remain connected to its contributors, institutions, funders, data, references, versions, and subsequent corrections. Those relationships must travel beyond the publisher’s website. Crossref and DataCite’s 2026 guide identifies metadata that helps people assess research integrity, including contributor roles, funding, versioning, retractions, licences, and links among research outputs. Publishers should improve both the completeness of their records and the speed at which updates reach discovery tools.
6. Make open access easier to navigate
Authors should not have to decipher conflicting instructions about funder requirements, rights retention, repository deposits, and publication charges. Publishers, libraries, and research offices can create clearer guidance at the point of submission and check compliance before an article is published. Open science also encompasses the research process: the Center for Open Science supports practices that make work more open, rigorous, and reproducible. Better compliance should reduce administrative effort while helping others examine and build on research.
7. Build economics that support quality and participation
We need to understand the combined cost of reading, publishing, preserving, checking integrity, and operating research infrastructure. Subscription, open access, institutional support, and AI licensing models should be judged by whom they serve, what they sustain, and whom they leave out. A new source of revenue is useful when it helps fund quality and broad participation. Publishers should be able to explain how new commercial models benefit authors, readers, societies, and the institutions that support research.
8. Design accessibility into production
Accessibility belongs in editorial and production specifications from the outset. Articles, figures, tables, video, and platforms should work for people with different abilities and on different devices. Publishers can make progress by assessing their existing collections, setting requirements for new content, and including accessibility in vendor selection. It is easier to produce an accessible article once than to repair many versions after publication.
9. Protect privacy and preserve the record
Research platforms collect information about what people read, search, and use. Organizations should set clear limits on data collection and sharing, assess security practices, and prepare for service interruptions. They should also ensure that scholarly content remains available when products, vendors, or business models change. Privacy, cybersecurity, and preservation require distinct plans, but all three affect whether researchers can rely on an information service over time.
10. Agree on outcomes that matter
Publication counts, downloads, and the number of AI features launched do not tell us enough about progress. By the end of 2026, each organization could choose a baseline and an accountable leader for a small set of measures: correction speed, metadata completeness, accessibility, compliance effort, verified reuse, affordability, or researcher experience. In 2027, we should report what improved and what did not. Research assessment initiatives such as CoARA and DORA offer useful perspectives on measuring contributions more responsibly.
Technology for a trusted scholarly corpus
These priorities depend in part on the quality of the underlying content. Gadget Software’s TopicLake™ Insights Engine offers an approach that transforms source material into structured, compute-ready documents (CRD). A January 2026 Signal65 White Paper examines TopicLake’s work with Federal Register material on Dell PowerEdge infrastructure with Broadcom networking. It describes an architecture for ingesting regulatory documents and making them queryable. Publishers should ask how well the approach preserves source context, rights, and the path from an answer back to evidence when applied to their own collections.
Other companies address different parts of that challenge. Data Conversion Laboratory works on converting content into structured formats, including JATS XML. Access Innovations offers controlled vocabularies and machine-assisted indexing. Silverchair’s Semedica enriches publisher content with semantic information. Impelsys describes turning content into computable, citable knowledge objects. SCITE adds citation context and reference checks, while Molecular Connections’ Tessera focuses on governing, attributing, and licensing publisher content for AI use. These are capabilities to examine through a publisher’s own test collection, with measurable checks for accuracy, discovery, provenance, accessibility, and rights.
Resources for the conversation
The following organisations offer standards, services, or examples relevant to these priorities. Listing them does not imply that they endorse this article.
| Priority | Organisations to explore |
| Responsible AI | STM; COPE; International Science Council |
| Research integrity | STM Integrity Hub; COPE; BioCore Trust Seal |
| Content rights and AI licensing | STM; Copyright Clearance Center; Creative Licensing International |
| Usable knowledge and metadata | Crossref; DataCite; ORCID |
| Open access and open science | cOAlition S; DOAJ; Center for Open Science |
| Sustainable participation | cOAlition S; DOAJ; Research4Life |
| Accessibility | W3C; DAISY Consortium; Fondazione LIA |
| Privacy and preservation | NISO; CLOCKSS; PORTICO |
| Broader measures of success | CoARA; DORA; STM |
Industry publications also help these conversations travel. Research Information, The Scholarly Kitchen, The Chronicle of Higher Education, The Geyser, David Worlock’s Coalition of the Curious, and Outsell bring different audiences and perspectives to questions about research, publishing, and the information economy. Mainstream outlets, including The New York Times and The Wall Street Journal, can help connect developments in scholarly communication to their wider public consequences.
The challenge I would have put to colleagues in Frankfurt is this: What will your organization change by December 2026, and what evidence will you share in 2027 that the change helped? Together we can advance humanity through quality research when people can find it, understand it, verify it, preserve it, and use it with confidence. I look forward to your feedback, ideas, suggestions, and action.
Have a great STM Frankfurt Dinner and Conference and the Frankfurter Buchmesse!
Darrell W. Gunter is Managing Director of Gunter Media Group