TL;DR: The International Committee of Medical Journal Editors (ICMJE) published an updated edition of its Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals in January 2026. Per ICMJE’s own summary of the revision, the update adds a new, standalone Section V, “Use of Artificial Intelligence in Publishing” (previously a single subsection under authorship), and a new Section III.L.2 addressing authors’ access to underlying data for reported research, sitting alongside a revised Section III.L.1 on clinical trial registration. Several other sections were also revised. This article summarizes what ICMJE itself says changed and links to its primary source; where ICMJE’s public pages did not give us enough detail to state a specific rule with confidence, we say so rather than guess.
Confirming this is real: what ICMJE’s own site shows
Because a claimed standards revision is exactly the kind of thing that’s easy to get wrong secondhand, this page is based on a direct check of ICMJE’s own site (icmje.org) rather than third-party summaries. As of this writing, icmje.org/recommendations displays “Updated January 2026” and ICMJE’s own news-and-editorials page for the release confirms the update adds two new sections and revises several existing ones. That page also notes ICMJE has published an annotated PDF marking every change for readers who want the exact line-by-line diff — the single best source for precise wording, which this article does not attempt to reproduce in full.
New Section V: AI now has its own top-level section
Previously, ICMJE’s guidance on generative AI use lived as a single subsection within the authorship section (II.A.4), introduced in the May 2023 update — see CASRAI’s ICMJE Generative AI Policy entry and the related AI co-authorship rejection (ICMJE 2023) entry for that original rule. The January 2026 update elevates and expands this into a full Section V, with three named subsections:
- V.A — Use of AI by Authors. Journals should require authors to disclose at submission whether AI-assisted technologies (large language models, chatbots, image generators, etc.) were used in producing the work, and to describe how the tool was used, in the cover letter and in the appropriate section of the manuscript. AI tools cannot be listed as authors, because they cannot take responsibility for the accuracy, integrity, and originality of the work — a requirement of authorship under ICMJE’s own criteria. Authors remain accountable for any AI-assisted content, including checking it for accuracy.
- V.B — Use of AI by Reviewers. Reviewers who use AI tools during peer review should disclose that use and, in particular, should not upload a submitted manuscript into an AI system where confidentiality of the unpublished work cannot be assured, absent the authors’ explicit permission.
- V.C — Editors’ Role in Ensuring Responsible Use of AI. Editors are expected to query authors on AI use, apply the same confidentiality expectation to their own and their staff’s use of AI tools on submitted manuscripts, and make a substantive judgment call between routine uses (e.g. spell-checking, grammar correction) and content generation, which is not treated as equivalent.
For research administrators and authors already following the 2023-era disclosure rule, the practical obligations described above are largely a continuation of that existing norm — the structural change is that AI guidance is now organized as its own section rather than a single authorship subsection, and reviewer- and editor-facing obligations are now spelled out explicitly rather than left implicit.
New Section III.L.2: authors’ access to data
ICMJE’s own summary of the update lists a new subsection, numbered III.L.2, under “Clinical Trials,” which it describes as addressing authors’ access to the data underlying reported results. This is a genuinely new numbered subsection, not a restatement of the pre-existing individual-participant-data (IPD) sharing statement requirement that has applied to trial manuscripts submitted after 1 July 2018 (see ICMJE’s clinical trials guidance for that separate, longer-standing policy). We were not able to retrieve the full text of III.L.2 from ICMJE’s public site at the time of writing to quote its exact scope or thresholds — readers who need the precise operative language, including who must attest to data access and how, should consult ICMJE’s own annotated PDF of the update rather than rely on a secondhand paraphrase.
The direction of the change is consistent with a broader trend already well established elsewhere in scholarly publishing: authorship criteria and journal data policies increasingly expect that at least some named author had direct access to the underlying dataset, not only to the analyzed results, as a check against reporting that cannot be independently defended. CASRAI’s ICMJE Authorship Criteria guide covers the existing four-part authorship test this new provision sits alongside.
Also revised: registration, misconduct, sponsor agreements, and funding disclosure
ICMJE’s summary of the release lists four further sections as revised, without itemizing the specific wording changes on its public pages:
- III.L.1, Clinical Trial Registration. ICMJE’s existing, longer-standing registration policy requires public, prospective registration of clinical trials (including device trials) before enrollment of the first participant. The January 2026 update revises this section, per ICMJE’s own change summary; we could not confirm the specific nature of that revision (e.g. any change to registration timing or scope) from ICMJE’s publicly accessible pages at the time of writing, so we are not asserting a specific “tightened” rule here without being able to point to the operative text. Check ICMJE’s annotated PDF for the exact change.
- II.B, Sponsor Agreements and Academic Freedom. Revised; this section covers researchers’ and editors’ independence from sponsor control over publication decisions.
- III.B, Scientific Misconduct. Revised; covers journals’ and editors’ handling of suspected misconduct.
- IV.3.A, Reporting of Funding Sources. Revised; covers how funding/support is disclosed in a manuscript.
We’re deliberately not speculating about the substance of these four revisions beyond what ICMJE itself states was touched, since doing so without the primary text would risk exactly the kind of secondhand inaccuracy this page is trying to avoid.
Why this matters for research administrators
ICMJE’s Recommendations are followed or referenced by thousands of biomedical and health-science journals well beyond the committee’s own member journals, which makes any revision to them a practical compliance question for research offices, authorship-dispute resolution, and manuscript submission checklists — not just an editorial-policy curiosity. Three concrete implications for institutions supporting biomedical authors:
- Update any internal authorship or AI-disclosure guidance that cites ICMJE’s old II.A.4 AI subsection to point to the new Section V instead.
- If your institution supports clinical trialists, flag the new III.L.2 data-access provision for review once ICMJE’s full text is available, particularly for multi-site or sponsor-funded trials where data custody arrangements are already a recurring point of friction.
- Don’t assume “ICMJE requires X” claims circulating informally are accurate for the four sections revised without public detail (III.L.1, II.B, III.B, IV.3.A) until you’ve checked the primary text yourself.
Frequently asked questions
Is there really a January 2026 ICMJE Recommendations update?
Yes. ICMJE’s own site (icmje.org/recommendations) is dated “Updated January 2026,” and ICMJE’s news-and-editorials page for the release confirms the specific sections added and revised.
What’s new in the AI section?
AI guidance moved from a single authorship subsection (II.A.4, dating to May 2023) into a standalone Section V with three parts covering authors, reviewers, and editors respectively. See the breakdown above.
Does ICMJE now require authors to have access to trial data?
ICMJE’s change summary lists a new Section III.L.2 on “authors’ data access,” but ICMJE’s public pages did not give us the operative text at time of writing. Confirm the exact requirement against ICMJE’s own annotated PDF before treating any specific interpretation as settled.
Did prospective trial registration requirements change?
ICMJE lists Section III.L.1 (clinical trial registration) as revised in this update, but we could not confirm the specific change from ICMJE’s public pages. Don’t rely on secondhand claims about what changed here — check ICMJE’s primary text.
Does this replace ICMJE’s 2018/2019 data-sharing-statement policy?
Not based on what we could confirm. ICMJE’s separate, longer-standing individual-participant-data sharing statement requirement (applying to trial manuscripts submitted after 1 July 2018, and to trials first enrolling on or after 1 January 2019) appears to remain a distinct policy from the new III.L.2 provision described in the January 2026 summary.
Related CASRAI resources
- ICMJE Generative AI Policy — the original 2023-era rule this update reorganizes
- AI co-authorship rejection (ICMJE 2023)
- ICMJE Authorship Criteria: The 4 Requirements
- CRediT vs ICMJE authorship
- Generative-AI disclosure statement
Sources: ICMJE Recommendations (icmje.org, dated January 2026); ICMJE’s own news-and-editorials summary of the January 2026 revision. This article reflects what ICMJE’s public pages state as of the verification date and will be updated if ICMJE publishes further detail on the sections we could not fully confirm.
