CASRAI Dictionary

Category: Perspectives

Opinion, argument, and field-shaping commentary on research-administration standards.

  • Generative AI Academic Integrity Policy 2026: Why University Rules Still Don’t Agree

    Generative AI academic integrity policy in 2026 remains fragmented: new peer-reviewed research from Springer and Cambridge University Press argues that universities cannot credibly enforce integrity standards while their own AI rules stay incoherent, even as searchers hunt for a mythical “30% rule” that does not exist in higher-education policy. Disclosure-threshold rules are proliferating faster than any shared standard — and convergence needs a common taxonomy, not more institution-specific thresholds.

    Academic integrity policy on generative AI is the set of institutional rules governing when, how, and whether students and researchers must disclose the use of AI tools in coursework, assessment, and scholarly output. As of mid-2026, no cross-institutional consensus exists on disclosure thresholds, detection reliability, or enforcement — only a widening patchwork of course-by-course and department-by-department rules.

    What does 2026 research actually say about AI and academic integrity?

    Two major 2026 publications converge on the same diagnosis, even though they approach it from different angles. Taylor et al., writing in Higher Education (Springer, 2026), conclude that universities cannot credibly enforce integrity standards in the age of AI without first ensuring coherence between their stated policies — a coherence that, in practice, rarely exists across a single institution’s own departments and courses.

    Gallant et al.’s Cambridge University Press Element, Academic Integrity in the Age of AI (2026), frames the same problem in sharper terms: generative AI “has rapidly and universally disrupted teaching, learning, and assessing with integrity.” Neither publication treats this as a temporary adjustment problem. Both treat it as a structural governance gap.

    That gap is not merely academic. A systematic literature review published in MDPI Information (Bittle et al., 2025) — now cited in well over 250 subsequent papers — found the evidence base on generative AI’s impact on academic integrity in higher education growing far faster than any agreed institutional response to it. The research volume has outpaced the policy convergence it was meant to inform.

    Why are disclosure-threshold policies multiplying instead of converging?

    University AI policy has moved past outright bans, but what replaced them is not one model — it is at least four, operating simultaneously, often within the same institution. The result is a patchwork where a rule that applies in one seminar room is void in the next.

    Policy model Example What it requires Enforcement
    Prohibition University of Cambridge, Faculty of History and Philosophy of Science AI may not be used as a source or quoted directly Academic misconduct procedure
    Disclosure-with-permission University of Kent; Solent University AI use permitted if declared, aligned with “fairness, transparency, accountability” Declaration checked at marking
    Course-level discretion Carnegie Mellon University Individual instructors set the rule per assignment, from total ban to full permission Devolved to instructor/department
    Integrated-tool model Emerging across STEM and data-science departments AI treated as a citable tool, akin to a calculator or search engine Attribution required, not a use-threshold

    The Office of the Independent Adjudicator for Higher Education, which handles student complaints across UK universities, notes that almost all AI-related complaints it receives come from students already subject to a misconduct procedure — evidence that disputed detection and inconsistent policy, not deliberate misuse, drive much of the caseload.

    Three structural forces keep these models from converging:

    • Decentralised governance — departments and individual instructors set their own rules, so no single institutional policy actually governs a student’s experience.
    • Detection unreliability — AI-detection tools produce enough false positives that no institution can safely anchor discipline to a single similarity or probability score.
    • A moving technical target — a policy calibrated to one model generation is frequently obsolete by the next; UNESCO-cited research highlights how generative AI is disrupting assessment methods that rely on final written output, such as essays, faster than institutions can rewrite their rules.

    Is there really a “30% rule” for AI use at university?

    No. Search interest in a “30% rule for AI” in academic-integrity contexts is real, but the rule itself is not an education-sector standard — it is a general AI-automation heuristic, describing a guideline that AI should handle roughly 70% of repetitive or preparatory work while humans retain the remaining 30% for oversight, creativity, and judgement in business and knowledge-work settings. No UK, US, or Australian university academic-integrity policy has adopted a codified 30% (or any single-number) disclosure threshold as of 2026.

    What some institutions have instead is a detection-review band: similarity or AI-probability scores that trigger human review of a submission, rather than an automatic misconduct finding. This is a procedural safeguard against detector false positives, not a permitted-use quota, and it varies by tool and by institution rather than following any shared figure. Searchers conflating the two are importing a business-automation concept into a governance vacuum that genuinely has no numeric answer yet.

    What would workable policy convergence actually require?

    A workable convergence needs a shared disclosure taxonomy, not another round of institution-specific thresholds. Three elements are prerequisites, based on where the Springer and Cambridge research locates the current failure points:

    • A common vocabulary for AI-use tiers — categories such as “AI-assisted drafting,” “AI-assisted research,” and “AI-generated content requiring full disclosure,” defined once and adopted consistently, rather than redefined by every syllabus.
    • Separation of detection from adjudication — using AI-detection scores only to flag cases for human review, never as standalone evidence of misconduct, addressing the false-positive problem identified in current casework.
    • Sector-level reference points, comparable to how research-integrity bodies such as COPE and ICMJE set shared expectations for publication ethics, giving individual universities a common baseline rather than each rebuilding policy from first principles.

    Institutional research-administration teams evaluating their own policy coherence can compare their current rules against a structured framework for research administration governance rather than treating AI-use policy as a standalone, one-off document.

    Common questions on AI and academic integrity

    Is using AI considered plagiarizing?

    It depends entirely on the institution’s specific policy and whether the use was disclosed. Using AI-generated content without proper attribution is treated as academic dishonesty at most universities, similar to unattributed use of another author’s work, but disclosed and permitted AI assistance is not automatically classed as plagiarism.

    What is the 30% rule for AI?

    The “30% rule” is a general AI-automation heuristic — AI handles roughly 70% of routine work, humans retain 30% for oversight and judgement — not an academic-integrity standard. No university has adopted a codified 30% disclosure or permitted-use threshold as of 2026; the term is being misapplied from business contexts into education searches.

    Can my university tell if I use AI?

    Sometimes, but not reliably. AI-detection tools can flag likely AI-generated text, and instructors often notice sudden shifts in writing style or fabricated citations, but detection software produces enough false positives that most institutions treat a flag as grounds for review, not automatic proof of misconduct.

    Is it plagiarizing if you use ChatGPT?

    It can be, depending on context and disclosure. Using ChatGPT-generated text without citation or acknowledgement is flagged as plagiarism under most current academic-integrity policies, while properly disclosed and permitted use — for example, brainstorming or editing assistance under a disclosure-with-permission model — typically is not.

    Implications for institutions, publishers, and standards bodies

    For research administrators, the near-term risk is reputational and legal, not just academic: enforcing a misconduct finding on an unreliable detector, against a policy a different department contradicts, is a weak position in an appeal — exactly the scenario the OIA’s casework note describes. Publishers and funders face an adjacent problem downstream, where undisclosed AI assistance in manuscript preparation raises the same coherence question long faced by human-authorship attribution: disclosure only functions as a standard when categories are shared, not improvised per venue.

    CASRAI originated the CRediT contributor role taxonomy in 2014 to solve a structurally similar problem — inconsistent, non-comparable attribution practices across journals. The standard is now stewarded by NISO as ANSI/NISO Z39.104-2022. AI-use disclosure in both teaching and research settings is heading toward the same fork: either a shared taxonomy emerges by deliberate convergence, or institutions continue absorbing the cost of policy fragmentation one appeal at a time.

    Until a sector body publishes a reference taxonomy for AI-use disclosure tiers, institutions should treat internal policy coherence — not a numeric threshold — as the actual compliance target for 2026.

  • Why Publish Open Access? A Case for Researchers, Funders and Institutions

    Why publish open access? Because immediate, paywall-free publication increases a paper’s readership and citation potential, satisfies funder mandates from cOAlition S, UKRI and Wellcome, keeps outputs REF-eligible, and extends publicly funded research to readers who cannot access subscription journals — benefits that typically outweigh the cost of article processing charges.

    Open access is a publishing model in which the final, peer-reviewed version of a research output is made freely available online at the point of publication, without a subscription or paywall, under a licence that permits reuse. That single design choice — removing the paywall — is what drives every benefit and every trade-off discussed below.

    Why does open access matter for visibility and citation?

    Removing a paywall expands a paper’s potential readership beyond subscribing institutions to independent scholars, clinicians, policymakers and researchers in lower-income countries. Publisher-commissioned meta-analyses report open-access citation advantages in the region of 18–40%, a range corroborated by Taylor & Francis and Springer Nature author-services data. Independent bibliometricians caution that part of this gap reflects self-selection — authors tend to pay for open access on papers they already judge to be their strongest — so the advantage should be read as a correlation, not a guaranteed multiplier.

    Visibility gains are strongest for interdisciplinary and applied fields, where readers outside a paper’s home discipline or sector are less likely to hold a subscription. For research administrators tracking impact, unrestricted access also improves the reliability of usage metrics reported to funders and REF impact case studies, since download and view counts are not artificially depressed by paywall friction.

    Why do funders require open access?

    A growing share of research funding now carries a binding open-access condition, not a recommendation. Non-compliance can mean ineligible outputs, clawed-back grant funds, or exclusion from future funding rounds — which is why open access has shifted from an ethical preference to a compliance requirement for most UK and European researchers.

    • cOAlition S / Plan S — launched in 2018, this consortium of research funders requires immediate open access with no embargo, typically under a CC BY licence, for all funded research articles.
    • UKRI — UKRI’s open access policy has applied to journal articles and conference proceedings from grants awarded since 1 April 2022, and extended to monographs, book chapters and edited collections from 1 January 2024.
    • Wellcome — Wellcome’s open access policy requires immediate open access under a CC BY licence for all research articles arising from Wellcome funding, with no embargo permitted.
    Funder Embargo permitted Preferred licence Route accepted
    cOAlition S (Plan S) No CC BY Gold or green with no embargo
    UKRI No (journal articles) CC BY Gold or green via repository deposit
    Wellcome No CC BY Gold, with preprint posting expected

    These mandates are the practical reason many researchers no longer treat open access as optional: the funding itself is now conditioned on it.

    Why is open access so expensive?

    The commonest objection to open access is cost. Gold open access is usually funded through an article processing charge (APC) paid by the author, institution or funder rather than by the reader, and APCs at established hybrid and fully open-access journals frequently run into several thousand pounds per article. That cost has not disappeared — it has moved from the reader’s library subscription to the author’s grant budget, which is precisely why the objection persists even as access improves.

    Three developments are making that cost more visible and, in places, avoidable:

    • Price transparency requirements — cOAlition S’s Price and Service Transparency Framework requires participating publishers to disclose a cost breakdown behind their APCs, rather than setting a single opaque list price.
    • Transformative and Read-and-Publish agreements — many UK institutions now hold deals with major publishers that bundle subscription and publishing costs, so individual authors at those institutions pay no APC directly.
    • No-fee routes exist — green open access (self-archiving the accepted manuscript in a repository) and diamond open access (journals that charge neither reader nor author) both avoid APCs entirely; a substantial share of journals indexed in the Directory of Open Access Journals charge no APC at all.

    The honest answer to “why is open access so expensive” is that the cost of publishing has not fallen — it has been reallocated and, under frameworks such as Plan S’s transparency requirement, made auditable in a way subscription pricing never was.

    Is open access REF-ready — and who else benefits?

    For UK institutions, open access is also an assessment-eligibility issue. REF’s open-access policy, first applied in REF 2021 and carried forward into REF 2029 preparations, requires eligible journal articles and conference proceedings to be deposited in an institutional or subject repository within three months of acceptance to count towards the exercise. An output published open access but deposited late, or not deposited at all, can be ruled ineligible regardless of its quality — making the deposit step, not just the publishing decision, the compliance-critical action.

    Beyond assessment mechanics, open access serves the public-benefit case that funders increasingly require research to articulate: publicly funded findings reaching the clinicians, teachers, small businesses, patient groups and policymakers who funded them through taxation but who never held a university library card. This is the same accountability logic behind open metadata and contributor-transparency standards more broadly. CASRAI originated the CRediT contributor role taxonomy in 2014 as one such standard; it is now stewarded by NISO as ANSI/NISO Z39.104-2022, and, like open access itself, exists to make the research record more usable to people beyond the original authorship team.

    Common questions about publishing open access

    Should you publish open access?

    In most cases, yes — and increasingly it is not discretionary. If your funder is part of cOAlition S, or is UKRI or Wellcome, open access is a condition of the grant. Even without a mandate, the visibility, compliance and public-benefit case generally outweighs the APC cost, particularly where a transformative agreement or green route removes that cost entirely.

    What are the benefits of open access publishing?

    The core benefits are wider readership, a documented (if contested) citation advantage, compliance with funder mandates, REF eligibility when deposited correctly, and public access for readers outside subscribing institutions. Authors publishing gold open access also typically retain copyright under a CC BY licence rather than assigning it to the publisher.

    Do authors pay for open access?

    Often, but not always. Gold open access is usually funded via an APC paid by the author’s institution or funder. Green open access (repository self-archiving) and diamond open access (no-fee journals) both let authors publish openly without paying an APC at all.

    What are the disadvantages of open access publishing?

    The main drawbacks are APC cost where no waiver or agreement applies, uneven journal-quality perceptions in some fields, and the administrative burden of tracking funder-specific licensing and deposit requirements. Predatory journals exploiting the APC model are a further, separate risk that authors should screen for via journal vetting tools.

    What this means for authors going forward

    The direction of travel is unambiguous: funder mandates are expanding, not retreating, and REF-style assessment exercises are tightening deposit compliance rather than relaxing it. Researchers, institutions and publishers who treat open access as a compliance and visibility strategy — choosing the route (gold, green or diamond) that matches their funder’s requirement and their budget — are better positioned than those who treat each publication decision in isolation. The cost objection remains real, but transparency frameworks and no-fee routes mean it is no longer the unanswerable objection it once was.

  • Research Assessment Reform: Why Collective Action Beats Solo Signatories

    Research assessment reform needs collective action because hiring, promotion and funding criteria are set independently by thousands of institutions — a single university dropping journal-based metrics gains nothing if every competing institution, funder and publisher still rewards them. Recent research-on-research literature frames this explicitly as a collective action problem: individual declarations such as DORA signal intent, but only coordinated, system-wide commitments — the model CoARA is built around — actually rewrite the incentives that determine careers.

    A collective action problem in research assessment is a situation where no single institution can achieve reform on its own without risking a competitive disadvantage, so change only happens when many actors move together under a shared, verifiable commitment.

    What Is the Collective Action Problem in Research Assessment Reform?

    A 2025 study in Minerva by sociologist Alexander Rushforth, “Research Assessment Reform as Collective Action Problem,” argues that research evaluation change cannot be reduced to individual institutional choice. Rushforth traces this through the Netherlands’ national “Recognition and Rewards” initiative, formally launched in 2019 to coordinate system-wide changes in assessment practice across the Dutch science system.

    The framing matters because it shifts the diagnosis. If assessment culture were simply a matter of institutional willpower, a DORA signature would be sufficient. If it is instead a coordination failure — where no actor can safely move first — then reform requires simultaneous, mutually reinforcing commitments from institutions, funders and publishers together.

    Why Doesn’t an Individual DORA Signature Change Hiring Criteria?

    The San Francisco Declaration on Research Assessment (DORA), launched in 2012, asks signatories to stop using journal-based metrics such as the Journal Impact Factor as a proxy for the quality of individual articles or researchers. Signing carries no binding enforcement mechanism, and DORA itself has long acknowledged that the harder work begins after signature — its 2019 guidance “You’ve signed DORA, now what?” explicitly frames hiring, promotion and funding criteria as the next, unfinished step.

    Two structural problems keep that step unfinished when institutions act alone:

    • First-mover risk. An institution that stops counting journal prestige in tenure review can be undercut in recruitment and rankings by peers who have not changed, because researcher CVs are still read against metric-based expectations elsewhere.
    • Interoperability failure. Where assessment criteria diverge sharply between institutions and countries, researcher mobility suffers — a candidate assessed holistically at one university may be filtered out by a metrics-based shortlist at the next.

    Neither problem is solved by any single signature. Both require peer institutions, funders and disciplinary societies to move on a broadly shared timetable.

    How Does CoARA’s Coordinated Model Differ From Individual Declarations?

    The Coalition for Advancing Research Assessment (CoARA) was formed around the Agreement on Reforming Research Assessment, which the European Commission signed and endorsed alongside DORA on 8 November 2022. Unlike a one-off declaration, CoARA requires member organisations to commit to a shared action plan with defined milestones, reported progress and working groups that develop common tools and criteria across institutions — moving assessment reform from individual pledge to managed, collective process.

    That coordination logic was reinforced on 4 December 2025, when CoARA and DORA released a joint statement on aligning their respective reform efforts rather than running parallel, uncoordinated campaigns. Science Europe’s April 2026 position statement, “Connecting Open Science and Research Assessment Reform,” makes the same point from the funder side: it treats open science and assessment reform as “mutually reinforcing and interdependent drivers of research cultures,” explicitly a multi-actor framing rather than an institution-by-institution one.

    Dimension Individual DORA signature Coordinated (CoARA-style) commitment
    Enforcement None — declaration of intent only Action plan with milestones and reporting
    Hiring/promotion criteria Left to each institution’s own timetable Shared working groups developing common criteria
    Competitive risk to first movers High — one institution changes alone Reduced — peers move on a shared timetable
    Researcher mobility Fragmented across institutions/countries Greater interoperability of criteria sought

    What Does the Dutch “Recognition and Rewards” Case Show?

    Rushforth’s analysis of Recognition and Rewards found that the initiative succeeded in uniting support from multiple influential national stakeholders — universities, funders and academic hospitals moving together — precisely because it was designed as a coordinated, system-wide commitment rather than a set of separate institutional pledges. It also documents genuine friction: critics raised concerns about the Netherlands “going it alone” internationally, illustrating that collective action problems exist at more than one level simultaneously — within a national system, and between that system and the rest of the world.

    The OECD’s April 2026 report “Reforming Research Assessment for Better Science” reaches a parallel conclusion at the international level, describing the current reform landscape as “a collective of organisations committed to reforming the assessment of research, researchers, and research organisations” — language that treats coordination, not individual compliance, as the operative unit of change.

    Frequently Asked Questions

    Does Signing DORA Actually Change University Hiring Practices?

    Not by itself. DORA’s own post-signature guidance states that hiring, promotion and funding decisions require separate, deliberate policy changes after signature. A signature is a public commitment; rewritten criteria documents, reviewed by hiring and promotion committees, are the actual evidence of change.

    What Is CoARA and How Does It Differ From DORA?

    CoARA is a coalition of research funders, institutions, and organisations built around the 2022 Agreement on Reforming Research Assessment. Unlike DORA’s single declaration, CoARA members commit to shared action plans, working groups and reported milestones — a coordination structure rather than a one-time pledge.

    Why Is Research Assessment Reform Described as a Collective Action Problem?

    Because no institution can safely change its own assessment criteria in isolation without risking a competitive disadvantage in recruitment and rankings. Research-on-research literature, including Rushforth’s 2025 Minerva study, argues reform requires simultaneous, coordinated commitments across many independent actors.

    Can One University Move Away From Metrics Without Being Disadvantaged?

    It can, but the Netherlands’ Recognition and Rewards case shows even a coordinated national effort faced criticism for “going it alone” relative to the rest of the world. A single institution acting without peer, funder and publisher alignment faces materially higher exposure to that same risk.

    What Should Institutions Actually Do Together?

    For research administration teams, the practical implication of the collective-action framing is direct: a DORA or CoARA signature belongs on a compliance checklist next to, not instead of, three coordination-dependent actions.

    1. Confirm hiring and promotion criteria documents have actually been rewritten, not merely a signature logged in a registry.
    2. Compare criteria against peer institutions in the same discipline and country to identify where first-mover risk is concentrated.
    3. Engage through CoARA working groups or equivalent sector bodies (ARMA, EARMA, INORMS) rather than drafting new criteria in isolation.

    Reform that stops at the signature stage produces a compliance artefact, not a changed incentive structure. The evidence from both the Dutch national case and the CoARA-DORA coordination model points the same way: assessment reform moves at the speed of the slowest coordinated group, not the fastest individual signatory. Institutions that treat their own criteria rewrite as contingent on parallel movement by peers, funders and publishers are following the pattern the research-on-research literature identifies as actually working — treating reform as a shared infrastructure problem, not a personal compliance decision.

  • CoARA Action Plan: Reform or Box-Ticking?

    CoARA’s action plan framework requires every signatory to publish, within a year of joining, a time-bound roadmap for reforming its research-assessment criteria, and to show progress at a five-year checkpoint due at the end of 2027. Three years after the Coalition’s November 2022 launch, membership has grown from roughly 100 founding organisations to more than 830 — yet CoARA’s own public tracker shows most signatories have not yet deposited a citable action plan, which is the real test of whether this is reform or box-ticking.

    The CoARA action plan is the documented, time-bound roadmap each Coalition for Advancing Research Assessment signatory must publish, setting out how it will revise the criteria, tools and processes it uses to evaluate research, researchers and research-performing organisations against the Agreement’s core commitments.

    What does the CoARA action plan actually require?

    The Agreement on Reforming Research Assessment (ARRA) obliges signatories to review or develop criteria, tools and processes against ten core commitments, and to record that process as an action plan with defined milestones. Under CoARA’s own guidance, the first plan is due within one year of signing (eighteen months for early signatories), with a further checkpoint at the end of 2027, by which point signatories must have completed at least one full review-and-development cycle.

    Crucially, CoARA imposes no fixed template. Organisations have “full freedom” in how they design their plan, and the Coalition explicitly asks signatories not to duplicate existing responsible-assessment work. That flexibility is defensible for a coalition spanning universities, funders, academies and research infrastructures — but it also means the Coalition has no standard unit for measuring whether commitments are being kept, only a request that plans be deposited publicly via a shared Zenodo collection.

    Has reform reached hiring, promotion and grant criteria?

    Some of the evidence is concrete. Loughborough University’s action plan, deposited in October 2023, embeds existing responsible research assessment practice into formal review criteria rather than treating CoARA as a new bolt-on process. Goldsmiths, University of London published a 2024–2029 plan explicitly tied to promotion and appraisal reform, and the University of Edinburgh deposited an updated plan in 2025 addressing how researchers and research-support staff are evaluated.

    Funders have moved too. Denmark’s Independent Research Fund (DFF) published an updated action plan in May 2025 that tracks delivery status against each commitment — a rare example of a signatory reporting progress rather than just intent. Italy’s national evaluation agency, ANVUR, has a 2024–2027 plan aimed at aligning national research-assessment criteria, not just one institution’s, with CoARA principles.

    These cases show the mechanism can produce real, checkable change in grant review and promotion documentation. The open question is how representative they are of the Coalition as a whole.

    How many signatories have actually filed an action plan?

    CoARA’s own live tracker — “Action Plans: Submitted & Pending to Date” — lists roughly 660 organisation entries with a due date for their first action plan. Of those, only around 136 carry an actual Zenodo DOI, meaning a plan has been deposited and made citable. The remaining entries, including many whose plans were originally due back in October 2023, are still marked “Pending” three years on.

    That is a completion rate of roughly one in five against CoARA’s own one-year deadline. It does not necessarily mean four in five signatories have done nothing internally — some may be reforming quietly without depositing paperwork — but it is the single clearest, most falsifiable indicator CoARA itself publishes, and it currently favours the “declaratory” reading of the Coalition’s progress over the “reformed” one.

    Metric Figure (CoARA live tracker, accessed July 2026)
    Organisation-level action plan entries tracked ~660
    Entries with a deposited, citable action plan (DOI issued) ~136 (≈21%)
    Entries still marked “Pending” ~514 (≈78%)
    Total current CoARA member organisations 834, across 60+ countries

    CoARA vs DORA: does history repeat itself?

    CoARA did not invent the credibility problem it now faces. The San Francisco Declaration on Research Assessment (DORA), launched in 2012 to curb inappropriate use of the Journal Impact Factor, has accumulated more than 27,000 individual and organisational signatures across 174 countries, according to sfdora.org’s own signer registry. Yet studies of research, promotion and tenure documents have repeatedly found continued reliance on journal-based metrics at institutions that formally signed DORA years earlier — a gap between signature and practice that critics now cite as the precedent CoARA risks repeating.

    CoARA’s design tries to close that gap by making the action plan, not the signature, the operative commitment, with a public deposit requirement and a 2027 checkpoint. A 2024 critique circulated on arXiv (Baccini et al.) argued the opposite risk: that shifting assessment toward qualitative, panel-based peer review could trade transparent metric-driven gatekeeping for a less transparent, harder-to-audit equivalent. Both critiques point to the same underlying test — not whether an organisation signs, but whether its actual review paperwork changes.

    Feature DORA (2012) CoARA (2022)
    Core ask Stop using Journal Impact Factor as a proxy for quality in funding, hiring and promotion Ten commitments on qualitative, diverse and open research assessment
    Accountability mechanism Voluntary signature; no mandatory public action plan Mandatory action plan within one year, deposited on Zenodo, checkpoint by end of 2027
    Current scale 27,000+ signatures, 174 countries (sfdora.org) 834 member organisations, 60+ countries (coara.org)
    Documented gap Continued JIF use found in signatory RPT criteria ~78% of due action-plan entries still “Pending” on CoARA’s own tracker

    Common questions about the CoARA action plan

    What is CoARA research?

    The Coalition for Advancing Research Assessment is a membership body of universities, funders, academies and research infrastructures committed to reforming how research, researchers and research-performing organisations are evaluated. It operates under the Agreement on Reforming Research Assessment, signed from November 2022, which sets shared commitments rather than a single enforced standard.

    What are CoARA National Chapters?

    CoARA National Chapters are country- or region-specific groups, such as the chapter for Ireland, that help local signatories interpret the Agreement’s commitments in their own funding, promotion and language context. They provide practical support for drafting action plans and coordinate national-level alignment with funder policy, including engagement with existing metrics guidance such as DORA.

    Is CoARA the same as DORA?

    No. DORA is a narrower 2012 declaration focused specifically on removing inappropriate Journal Impact Factor use from assessment. CoARA is a broader 2022 coalition with ten commitments covering qualitative assessment, output diversity and open science, and it requires a public, time-bound action plan rather than a one-off signature.

    How many organisations have signed CoARA?

    CoARA’s live membership register lists 834 organisations across more than 60 countries as of mid-2026, up from just over 100 at the November 2022 launch. Growth in membership has significantly outpaced growth in verified, publicly deposited action plans over the same period.

    What this means for research administrators

    For institutional leaders and research-administration teams, CoARA membership is not self-certifying reform. Signing the Agreement creates a public commitment; only a deposited, dated action plan against the ten commitments creates an auditable one. Institutions that have not yet filed should treat the gap as reputational exposure, not paperwork.

    • Check whether your organisation’s action plan (if due) has been deposited to the CoARA Zenodo collection, not just drafted internally.
    • Map each commitment against a specific, named change to hiring, promotion or grant-review criteria — not a general statement of intent.
    • Use the 2027 checkpoint as an internal deadline for demonstrating at least one completed review-and-development cycle, in line with the ARRA’s own timeframe.

    Outlook: what would count as proof by 2027?

    CoARA’s five-year touchpoint at the end of 2027 is the moment the “reform or box-ticking” question gets a real answer. If the proportion of signatories with a deposited, dated action plan rises substantially from today’s roughly one-in-five, and if more funders publish delivery-tracked updates in the style of Denmark’s DFF, the declaratory reading weakens. If the Pending column stays this full, CoARA will have reproduced the exact credibility gap DORA has spent over a decade trying to close.

  • Horizon Europe Proposal Won’t Survive at €175bn

    The European Commission’s €175 billion Horizon Europe proposal for 2028–2034 (FP10) is unlikely to survive Council negotiations intact. Every prior Multiannual Financial Framework (MFF) research settlement — including the current Horizon Europe programme, whose original €100 billion opening bid was cut to €95.5 billion — has been reduced during Council bargaining, and the Cyprus Council presidency has already tabled a lower figure. Research administrators building multi-year horizon europe proposal pipelines should plan around a materially smaller settlement, not the headline number.

    The Multiannual Financial Framework is the European Union’s seven-year budget ceiling, negotiated unanimously by the Council of the EU and agreed jointly with the European Parliament, within which programmes such as Horizon Europe and its successor, FP10, receive their funding envelope.

    What Does the €175 Billion FP10 Proposal Actually Contain?

    On 17 July 2025, the European Commission published its legislative proposal for FP10 — the tenth EU Framework Programme for Research and Innovation, running as “Horizon Europe” from 2028 to 2034. The headline figure is €175 billion, roughly double the €95.5 billion allocated to the outgoing 2021–2027 programme.

    That number sits inside a larger structure. According to the Commission’s own published breakdown, FP10’s €175 billion is nested within a €410 billion European Competitiveness Fund (ECF), alongside €234 billion for other schemes. The programme is organised into four pillars: Excellent Science (covering the European Research Council and Marie Skłodowska-Curie Actions), Competitiveness and Society, Innovation (the European Innovation Council), and a strengthened European Research Area pillar.

    Crucially, €175 billion is a Commission opening bid, not an agreed budget. Interinstitutional negotiation between Parliament, Council and Commission — the trilogue process — has only just begun, and a final MFF agreement is not expected before the end of 2026, ahead of the programme’s planned January 2028 launch.

    What Does MFF Precedent Say About Opening Bids?

    Every MFF research and innovation envelope in living memory has been negotiated down from the Commission’s opening figure. The pattern is consistent enough to build a forecast on.

    Framework programme Commission opening bid Outcome Change
    Horizon Europe (2021–2027) €100 billion (2018 Commission proposal) €95.5 billion final agreed budget, including a €5.4 billion NextGenerationEU top-up -4.5% net; at one point during the July 2020 European Council summit the figure was pushed as low as €80.9 billion before partial restoration
    FP10 / Horizon Europe (2028–2034) €175 billion (July 2025 Commission proposal) Not yet agreed. The Cyprus Council presidency has tabled €167.9 billion (April 2026) -4% on the Council’s opening counter-offer, against a European Parliament push for at least €200 billion

    The direction of travel is identical across both cycles: the Commission proposes a large increase, the European Parliament pushes for more, and the Council — which represents net-contributor member states with competing fiscal priorities — trims the figure during trilogue. FP10 is already following that script four months into formal negotiation, with the Cyprus presidency’s €167.9 billion counter-proposal landing before the Parliament has even finalised its own position.

    Why Is the €175bn Figure Already Shrinking?

    Three structural pressures point the same direction. First, the Council negotiates the overall MFF ceiling as a zero-sum allocation across cohesion, agriculture, defence and competitiveness spending — Horizon Europe/FP10 competes directly against those other headings, not in isolation. Second, several large net-contributor states have historically resisted MFF increases regardless of programme performance; this held even after Horizon Europe’s own interim evaluation found that every euro of EU contribution generates up to €11 in GDP gains by 2045 and that the programme had funded over 15,000 projects worth more than €43 billion as of January 2025.

    Third, FP10’s link to the European Competitiveness Fund creates a new negotiating lever that did not exist in the FP9 round: Council delegations can trade the research envelope against the wider €410 billion ECF total rather than negotiating Horizon Europe’s budget as a standalone line. That structural change makes a cut easier to justify politically, because ministers can present a smaller Horizon Europe figure as reallocation within a still-large competitiveness package rather than as a straightforward science-budget reduction.

    What Does This Mean for Grant-Pipeline Forecasting?

    Institutional research offices, EARMA and ARMA-affiliated grant teams, and funder relations units that are building multi-year FP10 pipeline models on the €175 billion figure are working from a number that has already moved once, before formal Council conclusions have even been reached. Practical implications include:

    • Model a range, not a point estimate. Use €167.9 billion (current Council presidency position) as a working floor and €175 billion as a ceiling until trilogue concludes, rather than planning around the Commission’s original figure.
    • Expect pillar-level reallocation, not uniform cuts. Past MFF rounds have shown cuts land unevenly across pillars; Excellent Science and EIC allocations have historically been better protected than collaborative-project envelopes.
    • Anticipate a later call-schedule start. With final agreement not expected before end-2026 and launch set for January 2028, first-wave FP10 call texts are likely to be finalised later in 2027 than institutions may be assuming.
    • Track the European Parliament position separately from the Council’s. The Parliament’s push for €200 billion is a genuine counterweight in trilogue, so the final figure could land above the Council’s current €167.9 billion offer — plan for a range, not a single downside scenario.

    For institutions coordinating this work through research administration functions, the practical response is to build FP10 revenue forecasts as scenario bands tied to the trilogue calendar, and to revisit those bands each time a Council presidency publishes a new negotiating box.

    Answer-First Q&A

    What Is the Budget for Horizon Europe?

    The outgoing Horizon Europe programme (2021–2027) has a final agreed budget of €95.5 billion, including a €5.4 billion NextGenerationEU top-up. The Commission has proposed €175 billion for its successor, FP10 (2028–2034), but that figure is an opening bid still subject to Council and Parliament negotiation.

    What Is the Budget of the Horizon Europe Pillars?

    FP10 is structured across four pillars: Excellent Science, Competitiveness and Society, Innovation, and the European Research Area. The Commission has not yet published final per-pillar allocations for FP10; these will be set through the same trilogue process determining the overall €175 billion headline figure.

    How Much Does the UK Pay Into Horizon Europe?

    The UK associated to Horizon Europe from January 2024 under a bespoke deal negotiated after the Windsor Framework, paying a contribution linked to UK GDP with a correction mechanism if UK entities draw significantly less funding back than they contribute. Exact annual figures are published periodically by UKRI rather than fixed in the framework regulation itself.

    What Is Horizon Europe Funding?

    Horizon Europe funding supports research and innovation projects across the EU and associated countries, covering frontier science (European Research Council), collaborative research addressing societal challenges, and innovation support (European Innovation Council). FP10 will continue this structure while adding closer integration with the European Competitiveness Fund.

    Conclusion: Plan for Less Than €175bn

    The evidence points one way. FP9’s opening bid fell by 4.5% net — and by nearly a fifth at its lowest negotiating point — before final agreement. FP10’s Council presidency has already tabled a 4% cut just months into formal talks, with a full trilogue still ahead. Research administrators, institutional finance offices and funder-relations teams should treat €175 billion as a ceiling, build FP10 grant-pipeline models around the €167.9–175 billion range the Council and Parliament are currently contesting, and revisit those forecasts as each successive Council presidency publishes its negotiating box through to the expected end-2026 agreement.

  • Is ORCID Legit? Its Limits Against Paper Mills

    ORCID is legitimate: it is a real, non-profit persistent-identifier registry used by thousands of publishers, funders and institutions worldwide. But “legitimate” is not the same as “fraud-proof.” ORCID’s own documentation confirms that an ORCID iD verifies control of an email address and account — not a researcher’s real-world identity, credentials, or institutional affiliation — which is exactly the gap paper mills exploit.

    An ORCID iD is a free, sixteen-digit persistent identifier that distinguishes one researcher from another and links that person to their publications, grants and affiliations. That single-sentence definition explains why ORCID is trusted — and why, on its own, it was never designed to stop organised authorship fraud.

    This piece is a CASRAI editorial perspective: it argues that identity-layer tools like ORCID and contribution-layer tools like the CRediT taxonomy solve different problems, and that conflating the two leaves a detection gap that paper mills are actively exploiting.

    Contents

    What does an ORCID iD actually verify?

    An ORCID iD confirms that a person controls a given ORCID account and email address, and it links that account to a persistent, disambiguated researcher record. It does not independently confirm a person’s legal identity, employment, qualifications, or that they actually authored the works attached to their profile.

    ORCID’s own guidance is explicit on this point: the organisation states that an ORCID iD is not a form of identity verification in the government-ID sense. Registration requires only a working email address, which is the same low bar that lets a legitimate early-career researcher register in seconds — and lets a paper mill spin up a disposable account just as fast.

    Is ORCID legit? The non-profit case

    Yes. ORCID is a genuine, mission-driven non-profit — ORCID is a global, not-for-profit organisation sustained by member fees from universities, publishers and funders, not by selling researcher data. It is embedded in submission workflows at major publishers and grant systems precisely because it solved a real problem: name ambiguity.

    • Common-surname collisions are severe — a mere hundred surnames account for over 85% of China’s population, with Wang, Li and Zhang alone covering more than a fifth, making name-only attribution unreliable at scale.
    • ORCID lets a researcher control which affiliations, works and peer-review activity are publicly visible, rather than relying on a publisher’s guesswork.
    • Adoption is now effectively mandatory at many funders and journals, which is a trust signal in itself — but mandated use is not the same as verified authenticity.

    So the trust question and the fraud-detection question are separate. ORCID earns trust as infrastructure; it was never marketed, and does not function, as an authorship-fraud filter.

    Can identity verification stop paper-mill authorship rings?

    No — not on its own, and the evidence for that is now substantial. The Committee on Publication Ethics (COPE) and the STM Association jointly defined and characterised the problem in their December 2022 Paper Mills Research Report, describing paper mills as commercial operations that manufacture fabricated or manipulated manuscripts and sell authorship slots, often to researchers under career pressure to publish.

    The scale became undeniable in the retraction data. Publishers retracted more than 10,000 research papers in 2023 — a record documented by Nature’s news team, with a large share traced to paper-mill-linked special issues, concentrated heavily at a single Wiley/Hindawi imprint before its special-issue programme was shut down. An ORCID iD attached to a fabricated paper did not prevent a single one of those retractions; in many cases the fraudulent authors held valid, active ORCID accounts throughout.

    That is the structural weakness: paper mills do not need to defeat ORCID, they only need to open an account, which requires nothing more than an email inbox. Publishers in response formed the STM Integrity Hub, launched in 2022, a shared cross-publisher infrastructure that pools signals — duplicate submissions, manipulated peer-review rings, image and reference manipulation — across member publishers in something ORCID’s single-account model cannot replicate, because ORCID has no mandate or mechanism to police manuscript content.

    Why a contribution taxonomy like CRediT tackles a different problem

    Identity tools answer “who is this person?” Contribution taxonomies answer a different, equally necessary question: “what did this specific person actually do on this specific paper?” CRediT, the contributor role taxonomy, was originated by CASRAI in 2014 and is now stewarded by NISO as ANSI/NISO Z39.104-2022 — CASRAI is the originator, not the current standards steward.

    CRediT requires each listed author to be tagged against a defined set of contributor roles — conceptualisation, data curation, formal analysis, writing, and others — for every submitted manuscript. That disclosure layer creates a different fraud signal than identity ever could: a co-authorship pattern where a name appears solely under “funding acquisition” across dozens of unrelated papers in a short window is implausible in a way that a valid ORCID iD alone will never flag, because ORCID has no visibility into role-level contribution claims.

    Neither tool substitutes for the other. Identity infrastructure and contribution disclosure address separate failure modes, and a detection strategy that leans on only one is structurally incomplete.

    Tool What it verifies or standardises What it does not do
    ORCID iD Persistent identifier; confirms control of an account/email and disambiguates a researcher’s name across works Does not verify legal identity, institutional affiliation or credentials
    CRediT taxonomy (ANSI/NISO Z39.104-2022) Standardises disclosure of which of the defined contributor roles each named author performed Does not verify that the named person exists, consented, or was even contacted
    STM Integrity Hub Shares cross-publisher fraud signals in real time — duplicate submissions, manipulated peer review, image reuse Does not itself confirm any individual author’s identity

    Answer-first Q&A

    Is ORCID legitimate?

    Yes. ORCID is a genuine, non-profit registry that provides persistent researcher identifiers and is integrated into submission systems at most major publishers and funders. Legitimacy as an organisation, however, is separate from its capacity to verify identity — ORCID confirms account control, not real-world credentials.

    Is it safe to share an ORCID iD?

    Yes. An ORCID iD is a public, non-sensitive identifier by design, and researchers control visibility settings for the underlying record. Sharing it on a CV, manuscript, or grant application does not expose private data, since ORCID does not store the kind of personal information used for identity theft.

    Should you use ORCID?

    Yes, for disambiguation and administrative efficiency. An ORCID iD saves time on grant and manuscript forms and reliably links a researcher’s outputs across name changes or institutional moves. It should not, however, be treated by editors or reviewers as evidence that a submission’s authorship is authentic.

    Is ORCID credible?

    Yes, as infrastructure. ORCID is a trusted, community-governed non-profit that cannot be bought by a commercial entity and does not sell researcher data. Credibility as a registry does not, however, extend to guaranteeing the integrity of any individual manuscript that cites an ORCID iD.

    Implications for publishers, institutions and funders

    Editorial offices that treat a valid ORCID iD as a clearance signal are relying on infrastructure built for a different job. A layered approach performs better:

    • Require CRediT contributor-role statements alongside ORCID iDs, so implausible role patterns become visible at submission, not after retraction.
    • Cross-check institutional email domains and affiliations, since paper mills routinely fabricate both.
    • Join or query shared infrastructure such as the STM Integrity Hub, which pools cross-publisher fraud signals ORCID was never designed to hold.
    • Treat ORCID account age and activity history as a weak signal only — a freshly created account attached to a first submission warrants closer editorial scrutiny, not automatic rejection.

    Conclusion: identity plus contribution, not identity alone

    ORCID is legitimate infrastructure doing exactly the job it was built for: disambiguating researcher identity across a fragmented publishing ecosystem. Expecting it to also police fabricated authorship asks a registry to perform forensic work it has no data or mandate to do. COPE and STM’s own analysis, and the 2023 retraction record, both point the same direction: stopping paper mills requires layered defences — identity infrastructure, authorship policy, contribution disclosure, and shared publisher intelligence — working together, because no single layer was designed to catch what the others miss.

  • Guest Authorship: Why CRediT Alone Fails

    Guest authorship occurs when a research paper’s byline, or its CRediT contributor statement, names someone who did not perform the work described. Because CRediT statements are self-reported by the corresponding author with no independent check, they can record a false contribution as easily as a true one — the taxonomy documents intent, not proof.

    Guest authorship is the practice of crediting an individual — typically an influential or senior figure — as an author or contributor on a study they did not substantively perform, in order to lend the paper credibility or satisfy a hierarchy. It sits alongside gift authorship (crediting a colleague as a favour) and ghost authorship (omitting someone who did the work), and all three predate CRediT by decades. The open question is whether a standardised contributor-role taxonomy actually closes the loophole, or simply gives guest authorship a more official-looking form to hide behind.

    What counts as guest authorship?

    The International Committee of Medical Journal Editors (ICMJE) sets four cumulative criteria for authorship: substantial contribution to the work’s conception or data; drafting or critical revision; final approval of the version published; and agreement to be accountable for it. A guest author fails at least the first criterion — and often all four — yet appears on the byline regardless.

    The Committee on Publication Ethics (COPE) defines a guest author as someone “added, with or without their knowledge, to make the author list look more impressive despite having no involvement with the research.” COPE’s authorship flowchart, last updated in 2024, groups guest authorship with gift and coercive authorship as related but distinct forms of the same underlying problem: a byline that does not reflect who actually did the work.

    Guest, gift and ghost authorship compared

    These terms are frequently used interchangeably, but the mechanism and the harm differ in each case.

    Practice What happens Typical driver CRediT interaction
    Guest authorship An influential outsider is named as author for prestige, with no involvement in the study Boosting perceived credibility or acceptance odds Roles are invented and attached retroactively to justify the byline
    Gift authorship A colleague, mentor or junior researcher is credited as a favour or reward Reciprocity, career support, maintaining relationships Minor or symbolic roles (e.g. “supervision”) are assigned regardless of actual input
    Coercive authorship A senior figure insists on inclusion because they run the lab or hold the funding Power imbalance between principal investigator and juniors The senior author dictates their own — and sometimes others’ — declared roles
    Ghost authorship Someone who did substantial work (often a medical writer) is omitted entirely Commercial sponsors wanting distance from the publication The omitted contributor’s real role never appears in the statement at all

    Why self-declared CRediT statements don’t stop it

    CASRAI originated the CRediT contributor role taxonomy in 2014; the standard is now stewarded by NISO as ANSI/NISO Z39.104-2022. It gives editors and readers 14 defined roles — from Conceptualization to Writing – Original Draft — in place of an undifferentiated author list. That is a genuine improvement in transparency. It is not, on its own, a verification system.

    Three structural gaps explain why:

    • No independent attestation. The corresponding author typically submits the entire CRediT statement on behalf of every co-author. Most journal workflows do not require each named contributor to individually confirm their assigned roles before publication.
    • No cross-check against evidence. A “Formal analysis” or “Investigation” tag is accepted as declared; journals do not routinely request lab notebooks, data-access logs or version-control history to substantiate it.
    • Power dynamics survive the paperwork. A principal investigator who insists on inclusion can equally insist on which role is recorded against their name. The taxonomy formalises the description of a contribution; it cannot compel the description to be honest.

    The scale of the underlying problem predates CRediT and has persisted through its adoption. A 2011 BMJ cross-sectional survey by Wislar and colleagues, examining high-impact medical journals that already required contributorship disclosure, found honorary authorship in 21% of sampled research articles — direct evidence that a disclosure requirement, by itself, does not eliminate the practice it is meant to surface. Ghostwriting scandals tell the same story from the other direction: the withdrawal of the diet drug dexfenfluramine (Redux) from the US market in 1997, after reports linking it to cardiac valve injury, followed years in which academic names had been attached to industry-drafted manuscripts on the drug’s safety — a pattern documented in subsequent publication-ethics literature on pharmaceutical ghostwriting.

    What would actually close the gap

    Closing the gap requires moving verification outside the self-reporting author group. Several mechanisms already exist in partial form and could be combined into a working check.

    • ORCID-linked contributor confirmation. ORCID iDs already let researchers verify affiliations and works against institutional records. Requiring each co-author to confirm their own CRediT roles via their ORCID account — rather than accepting a single submission from the corresponding author — would close the “submitted on your behalf” loophole.
    • Editor-level plausibility checks. COPE’s flowchart already lists warning signs — implausibly long author lists, late additions, unresponsive co-authors — that editorial staff can screen for before acceptance, without new infrastructure.
    • Publisher-side integrity screening. Cross-publisher initiatives such as the STM Integrity Hub, run by the International Association of STM Publishers, pool signals across journals to flag manuscripts and author patterns associated with paper mills and authorship manipulation, extending scrutiny beyond what any single journal can see alone.
    • Institutional sign-off at submission. Some research offices now require every named author to countersign the submitted contributor statement before a manuscript leaves the institution — shifting the accountability point upstream of the journal entirely.

    None of these is sufficient alone. Combined, they replace a single self-declared statement with several independent points where a false claim can be caught before publication rather than after retraction.

    Answer-first Q&A

    What is a guest authorship?

    Guest authorship is when an individual is named as an author or contributor on a research paper despite having made no substantive intellectual or practical contribution to the study. The name is typically added to lend prestige, improve perceived credibility with reviewers, or satisfy an informal hierarchy inside a research group.

    What is honorary guest authorship?

    Honorary guest authorship describes the same practice as gift authorship: crediting a senior or well-known researcher — often a department head or supervisor — who provided general oversight or facilities but did not meet formal authorship criteria such as those set by the ICMJE. It is one of the most commonly reported forms of authorship misconduct.

    What are the four problematic types of authorship?

    Publication-ethics literature groups authorship misconduct into guest, gift, coercive and ghost authorship. Guest and gift authorship credit someone who did not contribute; coercive authorship results from a power imbalance forcing inclusion; ghost authorship is the reverse — omitting a genuine contributor, often a paid medical writer, from the byline entirely.

    What does authorship mean?

    Under ICMJE criteria, authorship requires substantial contribution to a work’s conception or data, drafting or critical revision of the manuscript, final approval of the published version, and accountability for the work’s accuracy and integrity. All four conditions must be met; meeting only one does not qualify a contributor for the byline.

    Implications for editors, institutions and funders

    For editors, the practical implication is that a CRediT statement should be treated as a starting point for scrutiny, not a closing one. Plausibility checks already recommended by COPE cost nothing to implement and catch the crudest cases — implausible author counts, contributions that don’t match a co-author’s known expertise, late-stage additions to the byline.

    For institutions and funders, the implication is upstream: research integrity offices and grant terms can require ORCID-verified, individually confirmed contributor statements as a condition of institutional co-authorship or funding acknowledgement, rather than leaving verification entirely to journals with limited capacity to investigate.

    For developers building submission systems, the opportunity is to make individual confirmation the default workflow rather than an opt-in extra — turning a document that records one person’s account into one every named party must attest to.

    CRediT made contributorship visible. Making it verifiable is the unfinished half of the same reform, and it will require identity infrastructure and editorial process — not a taxonomy update — to complete.

  • FAIR Dataset Mandates Risk Becoming a Checkbox

    A FAIR dataset is one that meets the Findable, Accessible, Interoperable and Reusable principles published in Scientific Data in 2016 — but a funder mandate requiring deposit and a data management plan does not, on its own, guarantee this. Genuine FAIR compliance demands rich metadata, persistent identifiers and community-standard formats that most minimally compliant deposits skip entirely, because current incentive structures reward the act of depositing, not the work of curating.

    A FAIR dataset is a digital research object — data or its metadata — that satisfies the Findable, Accessible, Interoperable and Reusable principles first formalised by the FORCE11 community and published in Scientific Data in March 2016. The principles were designed to be applied in degrees, not as a pass/fail gate, which is precisely where funder policy and researcher practice have diverged.

    What does a FAIR dataset actually require?

    The FAIR principles set out four categories of requirement, each broken into specific sub-criteria. They are deliberately conceptual rather than prescriptive, which is a strength for cross-disciplinary adoption and a weakness for enforcement.

    • Findable — data and metadata carry a globally unique, persistent identifier and are indexed in a searchable resource.
    • Accessible — retrieval uses a standardised, open protocol, with metadata remaining accessible even when the underlying data cannot be.
    • Interoperable — data and metadata use a shared, formal language and vocabularies that follow FAIR principles themselves.
    • Reusable — data carry a clear licence, detailed provenance, and conform to domain-relevant community standards.

    The Research Data Alliance’s FAIR Data Maturity Model, published in 2020, decomposes these four principles into 41 discrete indicators covering both data and metadata. That granularity matters: a dataset can satisfy some indicators and fail most others while still being described, loosely, as “FAIR.” A funder checking only for repository deposit is verifying perhaps one or two of the 41.

    Why do funder mandates default to minimal compliance?

    Funder FAIR requirements typically operationalise as two things: a submitted data management plan and a deposit in a recognised repository at the end of the project. Neither step audits metadata richness, vocabulary use, or licensing clarity. The result is a policy that is easy to comply with and easy to satisfy without producing a dataset anyone outside the original team could actually reuse.

    Three structural gaps explain why:

    • Resourcing. Science Europe’s funders’ briefing on data management planning recommends that compliant curation cost roughly 5% of total research budget — a figure rarely built into grant awards, leaving curation as unfunded overhead.
    • Recognition. Data curation is not weighted in hiring, promotion or tenure decisions in most institutions, so time spent enriching metadata competes directly with time spent on publications that do count.
    • Standards gaps. Many disciplines still lack the domain-relevant community vocabularies that Interoperability and Reusability depend on, so even willing depositors have nothing FAIR-compliant to conform to.

    Horizon Europe requires that all data produced under the programme be FAIR “by default,” which is the strongest funder-level statement of intent currently in force. Yet the European Commission’s own guidance materials acknowledge that FAIRness is a spectrum, not a binary condition — an admission that sits uneasily alongside a compliance model built around a single deposit checkpoint.

    The maturity gap: from “FAIR start” to genuine reusability

    The European Commission’s Joint Research Centre published FAIR Data Guidelines in 2025 that organise the RDA’s 41 indicators into five progressive maturity levels. The framework is useful precisely because it makes visible how far “minimally compliant” sits from “genuinely reusable.”

    Maturity level What it requires
    FAIR start Published in a catalogue with mandatory metadata; data itself is not structured for machine reuse.
    FAIR play Links added between datasets and related resources, with enriched provenance and cross-referencing.
    FAIR go Data structured to community standards, with defined terminologies (not necessarily machine-readable).
    FAIR share Machine-readable data models (JSON Schema, XML Schema, SHACL) with richly documented provenance.
    FAIRest of them all Machine-readable model endorsed by the domain community; terms exposed via shared FAIR vocabularies.

    Most mandate-driven deposits land at “FAIR start” — indexed, licensed, discoverable, but not structured for genuine machine or cross-team reuse. The JRC guidelines are explicit that not every dataset needs the top tier, but they are equally explicit that FAIRness can degrade over time if metadata and platforms are not actively maintained. A one-off deposit satisfying a funder’s closeout requirement is not maintenance; it is a snapshot.

    Rebuilding incentives for genuine data stewardship

    Treating FAIR as a compliance checkbox is a governance failure, not a researcher failure. Three changes would shift the incentive structure toward genuine stewardship rather than deposit-and-forget behaviour.

    1. Credit the labour. CASRAI originated the CRediT contributor role taxonomy in 2014, and the standard is now stewarded by NISO as ANSI/NISO Z39.104-2022. “Data curation” is one of its fourteen roles, offering institutions an existing, citable mechanism to formally recognise stewardship work in author contribution statements — a mechanism that remains inconsistently applied in promotion and tenure review.
    2. Fund it explicitly. Grant budgets should ring-fence curation costs at the level Science Europe’s own guidance recommends, rather than treating data management plans as an unfunded compliance document.
    3. Audit maturity, not deposit. Funders and institutions should reference maturity models such as the RDA’s 41 indicators or the JRC’s five-level scale in closeout review, rather than accepting repository deposit as sufficient evidence of FAIR compliance.

    FAIR is also not a complete governance answer on its own. The CARE Principles for Indigenous Data Governance, released by the Global Indigenous Data Alliance in 2019, extend the framework to cover collective benefit, authority to control, responsibility and ethics — dimensions that a pure findability-and-format checklist does not touch. Institutions building data policy around FAIR alone are optimising for machine reuse while leaving governance and consent questions unaddressed.

    Frequently asked questions

    What is a FAIR dataset?

    A FAIR dataset satisfies the Findable, Accessible, Interoperable and Reusable principles published in Scientific Data in 2016. It carries a persistent identifier, standardised access, shared vocabularies, and clear licensing and provenance — not merely a repository listing.

    What does FAIR stand for with data?

    FAIR stands for Findable, Accessible, Interoperable and Reusable. The acronym describes a framework for data stewardship, not a certification; the Research Data Alliance breaks it into 41 measurable indicators rather than a single pass condition.

    What does FAIR stand for in data management?

    In data management, FAIR describes the target state a data management plan should work toward: identifiers, rich metadata, open protocols and community-standard formats. It guides curation decisions throughout a project, not just the final deposit.

    Why does FAIR data matter?

    FAIR data matters because it lets both humans and machines discover, verify and reuse research outputs without contacting the original authors. Poorly curated “FAIR” deposits undermine reproducibility and waste the public investment funders intended the mandate to protect.

    Implications and outlook

    Funder FAIR mandates have succeeded in one respect: deposit rates have risen sharply since 2016. They have not, on current evidence, produced datasets that are reliably machine-actionable or cross-team reusable at scale. That gap will not close through stricter wording in policy documents; it requires funders to resource curation at realistic cost, institutions to credit it in career progression via mechanisms such as CRediT’s Data curation role, and disciplines to finish building the community standards that Interoperability depends on. Until those three conditions are met, “FAIR by default” will remain a policy aspiration rather than a description of the average deposited dataset.

  • Research Misconduct Examples: Why Reports Hide

    Research misconduct examples are well documented in scholarly literature and case databases, but the investigation reports that actually establish them almost never reach the public. In the UK and the US, institutions overwhelmingly keep findings, evidence and reasoning confidential even after a case closes — a practice that a small number of national bodies elsewhere have abandoned in favour of publishing final rulings.

    Research misconduct is fabrication, falsification, or plagiarism in proposing, performing, reviewing, or reporting research, as defined by the US Office of Research Integrity under 42 CFR Part 93. That definition is narrow by design. It is also, this piece argues, too often the only thing the public is allowed to see — the “what” without the “how we know.”

    What is research misconduct? Definitions and examples

    Fabrication, falsification and plagiarism — FFP — form the internationally recognised core of research misconduct, used by the US Office of Research Integrity (ORI), the UK Research Integrity Office (UKRIO), and the Committee on Publication Ethics (COPE). Fabrication means inventing data or results outright. Falsification means manipulating materials, equipment, images or data so the record misrepresents what actually happened. Plagiarism means presenting someone else’s ideas, data or words as one’s own without attribution.

    Beyond the FFP core, most UK institutional policies and UKRIO’s own guidance extend coverage to related breaches that damage the research record without always meeting the strict fabrication-falsification-plagiarism test:

    • Undisclosed conflicts of interest that could plausibly bias findings or peer review
    • Inappropriate authorship — denying credit to real contributors or granting it to non-contributors, a problem CASRAI’s contributor-role work in CRediT was originally built to address
    • Redundant or “salami-sliced” publication of the same dataset across multiple papers
    • Manipulation of the peer-review process, including fabricated reviewer identities
    • Paper-mill involvement — buying or selling fraudulent manuscripts dressed as original research

    Retraction Watch’s public database tracks the downstream signal of these failures — the retraction itself — but a retraction notice rarely explains what an investigation found. That gap between “a paper was retracted” and “here is the evidence” is the transparency problem this article addresses.

    Why do institutions keep investigation reports confidential?

    UK and US institutions cite broadly consistent reasons for withholding full reports, and some are genuinely legitimate. None justifies blanket non-disclosure once a case is closed and a finding is made.

    • Personnel-record status. Many US institutions classify misconduct files as personnel records, exempting them from public-records requests even at public universities.
    • Whistleblower protection. Confidentiality during an inquiry is defensible — it protects the person who raised the concern from retaliation while facts are established.
    • Reputational risk to the accused. Institutions worry that publishing findings, even redacted ones, will follow a researcher regardless of outcome.
    • Litigation exposure. Legal teams treat investigation reports as liability documents first and research-integrity documents second.
    • No statutory disclosure duty. Unlike the US, where ORI’s federal oversight of Public Health Service-funded research creates a disclosure mechanism, the UK has no national body with investigatory or publication powers.

    The UK’s 2019 Concordat to Support Research Integrity, coordinated by Universities UK, commits signatory institutions to publish an annual statement on how many misconduct allegations they received and investigated. It does not require publication of the underlying reports. This is the crux of the transparency debate: aggregate counts satisfy the letter of accountability while the substance — what was actually found, and how — stays locked in a drawer.

    The US system looks more transparent at first glance because ORI’s Case Summaries register is genuinely public. Read the fine print, though, and the scope narrows sharply: it covers only Public Health Service-funded research, only cases where ORI itself made a finding (institutions handle the great majority internally, without ORI referral), and only researchers currently serving an active administrative sanction — historical entries are removed once the sanction period expires.

    How do more transparent regimes handle disclosure?

    A handful of national systems have made a different bet: that publishing final misconduct decisions, with personal data appropriately redacted, strengthens rather than undermines trust in the research system.

    Jurisdiction Body Investigation outcomes published? Legal basis
    Denmark Danish Committees on Research Misconduct (Nævnet for Videnskabelig Uredelighed) Yes — final decisions published, with case facts, on a rolling basis Danish Act on Research Misconduct (2017)
    Norway National Commission for the Investigation of Research Misconduct (Granskingsutvalget) Yes — decisions published since the commission became operational in 2022 Research Ethics Act (2017, amended)
    Netherlands National Board for Research Integrity (LOWI) Advisory opinions published in anonymised form Netherlands Code of Conduct for Research Integrity
    United States Office of Research Integrity (ORI) Partial — case summaries for PHS-funded, ORI-adjudicated findings only 42 CFR Part 93
    United Kingdom None national; institutions decide individually Rare — annual aggregate statistics only, per the Concordat to Support Research Integrity No statutory research-misconduct framework

    Denmark and Norway share a structural feature the UK and US both lack: a single national committee with statutory authority to investigate and to publish. That centralisation removes the conflict of interest built into the UK and US model, where the institution investigating its own researcher is also the institution deciding whether the findings ever see daylight.

    What does the secrecy cost — and what should change?

    Withholding investigation reports has three concrete costs. First, it prevents the research community from learning the specific mechanics of a case — which is precisely what field-level prevention requires. A retraction notice reading “concerns about data integrity” teaches almost nothing; a published finding detailing exactly which images were duplicated, or which patient records were invented, teaches a field how the fraud was constructed and how it was caught.

    Second, it fuels distrust in the institutions research administrators run. When a case surfaces only through journalism or a whistleblower’s own account — rather than the institution’s — the institution looks like it concealed wrongdoing rather than corrected it, whatever the reality.

    Third, it weakens deterrence. Sanctions that never become public carry a materially smaller reputational cost, changing the incentive calculation for researchers weighing whether to cut corners.

    None of this requires abandoning legitimate protections. A workable model — closer to Denmark’s than the UK’s current default — would:

    1. Publish final findings only, after due process concludes, never mid-investigation
    2. Redact personal data unrelated to the finding itself, while naming the researcher where a finding of misconduct is confirmed and sanctioned, consistent with COPE’s retraction guidelines
    3. Separate the publication decision from the investigating institution, ideally through a national or sector body — the role UKRIO could grow into if given statutory footing
    4. Require research integrity offices to log outcomes in a format compatible with existing registries, including Retraction Watch’s database and journals’ own correction records

    UKRIO’s current remit is advisory, not investigatory — it cannot compel disclosure. Extending it, or creating a UK equivalent of Denmark’s committee, would close the gap between the UK’s stated commitment to research integrity and its practice of keeping the evidence for that commitment confidential.

    Common questions on research misconduct

    What counts as research misconduct?

    Research misconduct is fabrication, falsification, or plagiarism in proposing, performing, reviewing, or reporting research. It excludes honest error and genuine differences of scientific opinion. UK guidance from the UK Research Integrity Office also treats undisclosed conflicts of interest and data manipulation as misconduct.

    What are the three types of research misconduct?

    The three internationally recognised types are fabrication (inventing data), falsification (manipulating materials or altering results), and plagiarism (presenting others’ work as one’s own). This FFP framework, used by ORI and COPE, deliberately excludes honest error and normal scientific disagreement.

    What is considered the most serious form of research misconduct?

    Fabrication is generally treated as the most severe form because it invents an entire dataset or result rather than distorting a real one. In practice, sanctions are often harshest for falsification in clinical or biomedical research, where fabricated or altered findings carry direct patient-safety consequences.

    Investigation-report secrecy is a policy choice, not a technical necessity. Denmark and Norway show that publication and due process are compatible; the UK’s Concordat shows that aggregate transparency, absent case-level disclosure, is not the same thing as accountability. Research administrators and research integrity offices weighing their own disclosure policy now have a working alternative model to point to, not just a set of reasons to say no.

  • Is Plan S Open Access Working? A Sceptic’s Case for Differentiated Mandates

    Five years on from its 1 January 2021 compliance deadline, Plan S open access policy sits in an odd position: widely credited with putting open access on every funder’s agenda, yet quietly walked back by the very coalition that wrote it. An independent October 2024 review, Galvanising the Open Access Community: A Study on the Impact of Plan S, found the policy had a “game-changing” effect through its Rights Retention Strategy. But cOAlition S’s own 2026–2030 strategic plan tells a second story — one of phased retreat from the rigid, one-size-fits-all mandate it launched in 2018. That gap between celebratory retrospective and quiet course-correction is the real story, and it is worth asking plainly whether a blanket mandate was ever the right instrument.

    What Plan S Actually Requires

    Plan S was launched in September 2018 by cOAlition S, an international consortium of national research funders and charitable foundations that includes UKRI and the Wellcome Trust. Its ten founding principles required that, from 2021, all peer-reviewed publications resulting from funding by coalition members be made immediately open access — either in a fully open access journal or platform, or via deposit in an open repository with no embargo.

    Two mechanisms did the heavy lifting:

    • The Rights Retention Strategy (RRS), which lets funded authors apply a CC BY licence to their author-accepted manuscript regardless of the publisher’s own policy, enabling immediate green open access.
    • Article processing charges (APCs), the fee-based gold open access route, which cOAlition S initially agreed to fund on authors’ behalf where a compliant venue existed.

    Notably, the original ten principles were scoped to peer-reviewed journal articles and conference proceedings. cOAlition S explicitly deferred a firm mandate for monographs and book chapters, citing the different funding cycles, peer-review norms, and licensing conventions of humanities and social-science (HSS) publishing — an early acknowledgement that a single rulebook does not fit every discipline.

    The Case Against the Blanket Mandate

    The criticisms of Plan S are not new, but they have hardened rather than faded. Three stand out.

    Cost-shifting to APCs. By pushing gold open access as the default compliant route, Plan S moved the cost of publishing from reader-side subscriptions to author-side fees. Well-resourced institutions and grant-rich disciplines absorb this easily; early-career researchers, unfunded scholars, and institutions in lower-income countries do not. Critics — including Science (AAAS), in its 2024 “mixed review” of the policy — have argued this risks a pay-to-publish stratification that Plan S was meant to dismantle, not recreate.

    Disciplinary disparities. STEM fields, with large grant budgets and a journal-article-centred publishing culture, adapted to Plan S’s timelines relatively smoothly. Fields with smaller grants, more diffuse funding, or monograph- and edited-volume-centred outputs did not. A mandate calibrated to biomedical and physical-science funding flows does not transfer cleanly to a discipline where the primary scholarly output is a single-author book written over several years.

    The humanities and monograph fit problem. Books remain the primary currency of career advancement in much of the humanities. Open access book publishing carries different cost structures (often higher per-unit costs than a journal article), different licensing sensitivities (image rights, third-party permissions, translated quotations), and a much thinner diamond and institutional-press ecosystem to absorb the volume. Applying a journal-shaped policy to a book-shaped discipline was, on the evidence of cOAlition S’s own deferred treatment of monographs, recognised as a mismatch from the outset — yet the underlying tension has never been fully resolved.

    Open access route How it works Typical discipline fit Cost burden cOAlition S’s current stance
    Gold (APC) Author or funder pays a publication fee for immediate open access STEM, grant-funded fields Shifted to authors/funders Supported, but flagged as unsustainable at scale
    Green (repository, via RRS) Author-accepted manuscript deposited under a retained CC BY licence Broad, including HSS Low direct cost Core mechanism, actively promoted
    Diamond (no author or reader fees) Community- or institution-funded journals/platforms Broad, especially HSS and society publishing Institutional/consortial funding Increasing emphasis in the 2026–2030 strategy
    Transformative agreements Institutions pay combined subscription-plus-publishing deals STEM-heavy, large-consortium markets High, opaque Support being phased out

    cOAlition S’s Own Retreat From Rigidity

    What makes the sceptic’s case harder to dismiss is that cOAlition S has, in effect, conceded much of it. The coalition’s published strategy for 2026–2030 signals a deliberate shift away from the rigid instruments of the 2018 launch:

    • Support for transformative agreements — once framed as a transitional bridge to full open access — is being wound down, an implicit admission that offsetting deals entrenched incumbent publishers’ revenue rather than transforming the market.
    • The strategy explicitly states that “no single model can meet all needs”, formally endorsing a plurality of routes (green, diamond, community-owned platforms) instead of privileging APC-funded gold.
    • Diamond open access — non-APC, non-subscription publishing typically funded by consortia, learned societies, or institutions — receives markedly more strategic weight than it did in 2018, partly because it fits humanities and society-publishing contexts that APC-gold never did.
    • Implementation timelines and compliance routes have been extended and softened repeatedly since 2021, a pattern of flexibility that was largely absent from the original ten principles.

    None of this is framed by cOAlition S as a repudiation of Plan S. But read against the criticisms above, it is difficult to interpret the 2026–2030 strategy as anything other than a coalition adjusting a blanket mandate toward the differentiated approach critics have been requesting since 2018.

    Common Questions About Plan S

    What is Plan S in open access?

    Plan S is an open access mandate launched in 2018 by cOAlition S, a coalition of national research funders including UKRI and the Wellcome Trust. It requires that peer-reviewed outputs from coalition-funded research be made immediately open access on publication, either through a compliant journal or platform, or via a no-embargo repository deposit.

    Do I have to pay for open access?

    Not necessarily. Gold open access typically involves an article processing charge (APC) paid by the author or funder. Green open access via repository deposit and diamond open access (no author or reader fees) are both compliant, fee-free alternatives that Plan S — and increasingly cOAlition S’s own strategy — actively supports.

    Toward Differentiated Funder Mandates

    The evidence points toward a specific policy design failure rather than a failure of open access as a goal. A single compliance clock, a single funding assumption, and a single default route (APC-gold) were applied across disciplines with radically different publishing economies. The fix is not to abandon open access mandates but to differentiate them:

    • Route-neutral compliance that treats green, diamond, and gold as equally valid by default, rather than gold-as-default with green as an exception.
    • Discipline-aware timelines, recognising that a monograph-based field cannot realistically match a journal-article field’s production cycle.
    • Direct funding for diamond infrastructure in HSS fields, rather than expecting APC markets to develop where publishing economics do not support them.
    • Transparent reporting on cost-shifting, so funders and institutions can see whether a mandate is redistributing cost fairly or simply moving it from library budgets to grant budgets.

    For research administration teams managing funder compliance day to day, this is not an abstract debate — differentiated mandates mean different checklists, different budget lines, and different risk profiles by discipline, and institutional policy needs to reflect that variation rather than applying one open access rulebook across every faculty.

    Conclusion: What Should Come Next

    Plan S succeeded at the one thing a blanket mandate is good at: forcing the issue onto every funder’s and publisher’s agenda within a few years, where voluntary encouragement had achieved comparatively little in the preceding two decades. It failed, or at least strained badly, at the thing blanket mandates are structurally bad at — accommodating disciplinary and economic diversity. cOAlition S’s own 2026–2030 strategic pivot toward plural, discipline-flexible routes is the clearest evidence that the coalition has reached the same conclusion. The sensible reading is not “Plan S failed” or “Plan S succeeded”, but that the next generation of funder mandates should be designed as differentiated instruments from the outset, rather than retrofitted into flexibility five years after a rigid launch.