CASRAI Dictionary

Tag: NIH AI policy

  • AI Research, Innovation, and Accountability Act: What Research Offices Need to Know

    The AI Research, Innovation, and Accountability Act (AIRIA, S.3312) was a bipartisan US Senate bill that would have created federal risk tiers, transparency reporting, and certification duties for high-impact AI systems. It cleared the Senate Commerce Committee in July 2024 but died when the 118th Congress adjourned in January 2025. Its framework has not disappeared, however: near-identical risk-tier and disclosure ideas now surface in state AI statutes, in federal agency guidance, and in follow-on bills before the 119th Congress — several of which already touch how NIH and NSF handle AI in grant review.

    AIRIA is a defined legislative proposal, not a law currently in force: it is the bill that proposed classifying AI systems as “high-impact” or “critical-impact” and tasking the National Institute of Standards and Technology (NIST) with testing, evaluation, validation, and verification standards for the highest-risk category.

    What is the AI Research, Innovation, and Accountability Act?

    The AI Research, Innovation, and Accountability Act is a US Senate bill introduced on 15 November 2023 by Senators John Thune (R-SD), Amy Klobuchar (D-MN), Roger Wicker (R-MS), John Hickenlooper (D-CO), Shelley Moore Capito (R-WV), and Ben Ray Luján (D-NM). It proposed a risk-based federal framework rather than blanket rules for all AI.

    Core provisions included:

    • A two-tier risk classification for “high-impact” and “critical-impact” AI systems used in consequential decisions.
    • Mandatory transparency reports and risk assessments from developers and deployers of the highest-risk systems.
    • A NIST-led programme to develop testing, evaluation, validation, and verification (TEVV) standards.
    • A certification and enforcement structure housed at the Department of Commerce.
    • A consumer-education and industry working-group mandate to support voluntary compliance ahead of formal rules.

    Unlike the EU’s comprehensive AI Act, AIRIA targeted only the highest-risk use cases and left most research and low-risk commercial AI activity outside its scope.

    What happened to AIRIA in Congress?

    AIRIA advanced further than most AI bills of its era but still did not become law. The Senate Committee on Commerce, Science, and Transportation ordered it reported on 31 July 2024, and the Congressional Budget Office published a cost estimate on 6 December 2024. Under standard congressional procedure, any bill not enacted before a Congress ends is considered dead; AIRIA lapsed with the close of the 118th Congress on 3 January 2025 and was not carried forward automatically.

    That is not the end of the story. Several bills before the 119th Congress (2025–2026) reuse AIRIA’s building blocks — including the AI Accountability Act (H.R.1694), which directs a federal study of AI accountability measures, and the Future of Artificial Intelligence Innovation Act of 2026 (S.3952), which revives the NIST standards-and-evaluation mandate AIRIA proposed. None of these has replicated AIRIA in full, but the pattern is consistent: risk tiers, NIST-run testing standards, and disclosure duties keep reappearing in federal drafting, which is why the original bill remains a useful reference text even though it never passed.

    How does AIRIA interact with NIH and NSF grant compliance?

    AIRIA itself never reached the funding agencies, but the compliance gap it targeted — undisclosed or unaccountable AI use in high-stakes review processes — is already being filled through agency policy rather than statute. Research offices do not need AIRIA to pass to feel its logic in practice.

    • NIH issued NOT-OD-23-149, prohibiting NIH scientific peer reviewers from uploading grant application or critique content into generative AI tools, to protect peer-review confidentiality and integrity.
    • NSF issued a parallel notice on 14 December 2023 barring reviewers from entering proposal or review information into non-approved generative AI tools, with corresponding updates folded into the Proposal & Award Policies and Procedures Guide (PAPPG).
    • OMB Memorandum M-24-10, issued 28 March 2024, requires every CFO Act agency — including the parent departments of NIH and NSF — to designate a Chief AI Officer, convene an AI governance board, inventory AI use cases annually, and publish compliance plans.

    Research administrators should read AIRIA less as a future obligation and more as the missing statutory layer above rules that funders have already implemented administratively. If AIRIA-style provisions are eventually enacted, they would most plausibly formalise — not replace — the NIH and NSF confidentiality prohibitions and the OMB governance-board model that are already operating today.

    How does AIRIA compare with state AI laws and the EU AI Act?

    Research institutions rarely operate under one AI framework. Multi-state university systems, international co-investigators, and federally funded projects with EU partners are simultaneously exposed to federal inaction, an unsettled state landscape, and a phased EU regime.

    Framework Jurisdiction Status as of July 2026 Relevance to research offices
    AIRIA (S.3312) US federal (Senate) Died with the 118th Congress, 3 Jan 2025; ideas recur in newer bills Reference model for future federal risk-tier and disclosure rules
    OMB M-24-10 US federal (executive) In effect since 28 Mar 2024 Directly governs how NIH, NSF, and other agencies use AI internally
    NIH / NSF AI notices US federal agency policy In effect since Jun–Dec 2023 Bars generative AI use in peer review of grant applications
    Colorado AI Act (SB 24-205) US state Repealed by SB 26-189 (14 May 2026); never took effect Cautionary example — comprehensive state AI law can collapse before compliance deadlines
    Texas TRAIGA US state In effect 1 Jan 2026 Intent-based liability model; applies to any AI system touching Texas residents
    EU AI Act European Union Phased in Aug 2024–Aug 2026 Relevant to Horizon Europe co-investigators and EU-based research partners

    The Colorado reversal is the clearest recent signal: SB 24-205 was the first comprehensive US state AI law, but Colorado Governor Jared Polis signed its full replacement, SB 26-189, on 14 May 2026 — meaning the original statute never actually took effect. State AI law is moving fast and is not stable enough to treat any single statute as a durable compliance target.

    Common questions research administrators ask

    What is the AI Research, Innovation, and Accountability Act?

    It is a 2023 US Senate bill (S.3312) that proposed risk-tiered federal oversight of “high-impact” and “critical-impact” AI systems, including NIST-led testing standards and mandatory transparency reporting. It advanced through Senate Commerce Committee review in 2024 but was never enacted.

    What is the AI legislation situation in 2026?

    No single comprehensive federal AI statute exists in the United States as of mid-2026. Oversight instead comes from a patchwork of agency guidance (OMB M-24-10, NIH and NSF notices), a shifting set of state statutes (Texas TRAIGA in effect, Colorado’s law repealed and replaced), and several competing federal bills still in committee.

    What are the seven principles referenced in AI regulatory frameworks?

    Frameworks such as the EU AI Act commonly cite human agency and oversight, technical robustness and safety, privacy and data governance, transparency, non-discrimination and fairness, societal and environmental wellbeing, and accountability. AIRIA did not adopt this exact list but pursued the same accountability and transparency goals through US-specific risk tiers.

    Why research offices should track this now

    Waiting for a federal AI bill to pass before building internal AI-use policy is the wrong sequencing. NIH and NSF already enforce confidentiality rules on generative AI in peer review, OMB already requires agency AI governance boards, and state rules are changing faster than any single institution can absorb reactively — Colorado’s reversal took less than two years from enactment to repeal.

    Research offices should treat AIRIA as a design template, not a deadline. Institutions that map their existing AI-use disclosure practices against AIRIA’s risk-tier and TEVV concepts now will be positioned to adapt quickly if a successor bill — whether H.R.1694, S.3952, or a future proposal — advances further than AIRIA did. The direction of travel across federal agency guidance, state law, and the EU AI Act is consistent even where the US federal statute itself has stalled: more disclosure, more documented risk assessment, and more named institutional accountability for AI used in decisions that affect people’s funding, careers, and research records.

    For related compliance context, see CASRAI’s research administration resources and the CASRAI Dictionary for definitions of adjacent governance and compliance terms.

  • AI Grant Application Rules: A Compliance Checklist for Research Offices

    Research offices are fielding the same question from every principal investigator this cycle: what counts as acceptable AI grant application assistance, and what will get a proposal flagged? The honest answer is that funders have converged on a rough principle — AI can edit, but it cannot originate — while diverging sharply on enforcement, disclosure and consequences. Grammar-checking and language polishing with a large language model is now explicitly permitted almost everywhere. Using AI to draft the scientific argument, generate specific aims, or write an entire proposal is not, and that gap is where applications get rejected or, in NIH’s case, administratively withdrawn.

    This checklist reconciles the current rules from NIH, UKRI, the European Research Council (ERC) and NHMRC (with a note on NSF, since both funders publish closely watched AI guidance), and gives research administrators ready-to-adapt disclosure wording for applicants.

    AI-Assisted Editing vs AI-Generated Drafting

    Funder policies converge on a distinction between two categories of AI use, even where the exact wording differs.

    • AI-assisted editing: grammar and spelling correction, clarity and readability improvements to text the applicant has already written, translation, and administrative formatting. This is broadly permitted.
    • AI-generated drafting: producing the scientific rationale, specific aims, hypotheses, study design or an entire section without substantive human authorship. This is broadly prohibited, and in NIH’s case carries the risk of administrative withdrawal.

    Evaluation is treated as a separate, stricter category again. Every funder examined for this analysis — NIH, UKRI, ERC and NSF — bars peer reviewers from using generative AI to analyse, summarise or score applications, largely to protect the confidentiality of unpublished ideas.

    Funder Rules Compared: NIH, UKRI, ERC, NHMRC and NSF

    The table below summarises publicly stated positions as of mid-2026. Research offices should always check the current version of the cited policy, as several funders note their guidance will evolve.

    Funder Applicant drafting/editing use Full AI-generated content Disclosure required Peer reviewer AI use
    NIH Permitted for grammar, spelling and clarity only Prohibited; applications “substantially developed by AI” may be administratively withdrawn under NOT-OD-25-132 (effective 25 Sept 2025) No formal disclosure field; NIH uses AI-detection screening and caps most PIs at six applications/year Prohibited from using AI to analyse or critique applications
    UKRI Permitted for drafting, editing, idea generation and literature comparison Prohibited: applicants “must not use generative AI tools to generate an entire application, or sections of an application, without human involvement” Encouraged, not mandatory; disclosure does not affect assessment Prohibited from using generative AI in assessment
    ERC Permitted for brainstorming, literature searches, revising, translating and summarising Prohibited in substance: applicants retain “full and sole authorship responsibility”; text-similarity detection is used Not a separate mandatory statement Strict non-delegation policy: no AI summarising, assessing or draft-evaluation writing
    NHMRC Permitted for drafting, editing and organising ideas Applicant must verify accuracy against the Australian Code for the Responsible Conduct of Research; sensitive data must not enter public AI systems Not a separate mandatory statement Reviewers may use AI only to refine the wording of their own comments, not to evaluate or score
    NSF Permitted for proposal preparation assistance Proposers are responsible for accuracy of all content regardless of AI involvement Required: proposers must indicate the extent of generative AI use in the project description Reviewers barred from uploading proposal content to public AI tools (confidentiality breach)

    The Compliance Checklist for Research Offices

    Institutional research offices can use the following checklist when advising applicants ahead of submission.

    • Confirm which category the intended AI use falls into — editing/formatting versus content generation — before the applicant starts drafting.
    • Check the specific funder’s current AI policy page rather than relying on last year’s guidance; NIH, UKRI and NSF have all updated their positions since 2023.
    • Where disclosure is required (NSF) or encouraged (UKRI), draft the disclosure statement early and route it through the same sign-off as conflict-of-interest and human-subjects certifications.
    • Warn applicants against pasting unpublished proposal content, preliminary data, or collaborator information into free or public AI tools — this risks both confidentiality breaches and, in the EU/UK, data protection exposure.
    • Never advise applicants to use AI for peer-review-adjacent tasks such as scoring their own proposal in a way that substitutes for genuine self-assessment.
    • Keep a record of which AI tools were used and for what purpose, in case a funder requests it during a research-integrity enquiry.

    Common Questions on AI Use in Grant Applications

    Can I use ChatGPT to write my grant application?

    Most funders allow ChatGPT and similar tools for grammar checks, clarity edits and brainstorming, but not for drafting the scientific argument or specific aims. NIH, UKRI, ERC and NHMRC all place ultimate authorship responsibility on the applicant, so a proposal substantially generated by AI risks rejection.

    Does NIH allow AI-generated grant applications?

    No. Under NOT-OD-25-132, effective 25 September 2025, NIH treats applications or sections “substantially developed by AI” as not reflecting the applicants’ original ideas, and such submissions may be administratively withdrawn. NIH also screens for AI use and caps most principal investigators at six new or resubmitted applications per year.

    Do I need to disclose AI use in a grant application?

    It depends on the funder. NSF requires applicants to state the extent of generative AI use directly in the project description. UKRI encourages disclosure without penalty at assessment. ERC and NHMRC do not mandate a disclosure statement but still hold the applicant fully accountable for all AI-assisted content submitted.

    Can grant peer reviewers use AI to assess applications?

    Generally no. NIH, UKRI, ERC and NSF all prohibit reviewers from using generative AI to analyse, summarise or score proposals, largely to protect confidentiality and prevent unpublished ideas reaching public tools. NHMRC allows a narrow exception: reviewers may use AI only to polish the wording of their own comments.

    Template Disclosure Language for Applicants

    Research offices are repeatedly asked for standard wording applicants can adapt rather than draft from scratch. Two templates cover the main scenarios.

    Where disclosure is required or requested (NSF/UKRI-style):

    “Generative AI (tool: [name and version]) was used to [check grammar and clarity / generate an initial outline] of Sections [X]. All scientific content, analysis and conclusions are the original work of the named investigators, who take full responsibility for the accuracy and integrity of this application.”

    Where disclosure is not mandatory but institutions want a defensive record (ERC/NHMRC-style, kept on file):

    “The applicants used [tool name] to assist with editing and language clarity only. No AI tool was used to generate the scientific rationale, methodology, hypotheses or original data interpretation contained in this application.”

    Neither template substitutes for reading the specific solicitation text, which occasionally adds requirements beyond the funder’s general policy.

    Implications for Research Offices

    The practical challenge is that these policies are not converging on common language, so a one-size-fits-all institutional AI policy will misfire on at least one major funder. A UKRI-style permissive default with encouraged disclosure would not protect a PI from NIH’s administrative-withdrawal risk, and an NIH-style prohibition would leave NSF’s mandatory disclosure field unanswered.

    • Build funder-specific AI guidance into pre-award checklists rather than a single institution-wide statement.
    • Treat AI-use attestations the same way as financial conflict-of-interest disclosures — logged, dated and retrievable if a funder investigates later.
    • Extend research-integrity training to cover AI-specific risks: fabricated citations, hallucinated preliminary data, and inadvertent disclosure of unpublished ideas to public tools.
    • Coordinate with research administration leadership on how AI-use records intersect with existing misconduct and compliance processes.

    What Comes Next

    Evidence on outcomes is starting to complicate the compliance picture. A February 2026 Nature analysis found AI-drafted NIH proposals were more likely to be funded, but that funded proposals using AI assistance also tended to read more similarly to one another — a finding likely to sharpen funder scrutiny of homogenised language rather than loosen it. Expect NIH’s detection and application-limit measures to be tested over the next funding cycle, while UKRI, ERC and NHMRC continue to state their guidance will be revisited as the evidence base evolves. The safest institutional posture for now is documented, funder-specific caution: assume editing is safe, assume drafting is not, and keep a paper trail either way.

  • NIH’s Ban on AI-Generated Grant Applications: What NOT-OD-25-132 Requires

    The National Institutes of Health has put research offices on notice: proposals that look like they were written by a machine, and principal investigators who submit an implausible volume of them, will now be treated as a fairness problem rather than a productivity feature. NIH AI policy grant applications is the subject of NOT-OD-25-132, “Supporting Fairness and Originality in NIH Research Applications,” issued 17 July 2025 and effective for applications with due dates on or after 25 September 2025. The notice does two distinct things at once — it restricts the use of generative AI in drafting applications, and it caps the number of applications any one principal investigator (PI) can submit per year — and research administrators need to treat them as two separate compliance obligations, not one.

    What NOT-OD-25-132 Actually Says

    NIH’s position is narrower than “no AI allowed.” The notice states that applications must reflect the original ideas and scientific contributions of the investigators, and that NIH will not accept applications that are substantially developed by AI. That is a materiality threshold, not a blanket prohibition on tools — spell-checkers, reference managers, and light editorial assistance are not the target.

    The stated concern is specific: NIH flags plagiarism, fabricated or non-existent citations, and other misleading content generated by large language models as the practical harms it is trying to prevent, not AI use in the abstract. Applications judged to fall foul of this standard can trigger a research misconduct review and enforcement action, including referral to the Office of Research Integrity, disallowance of costs, or termination of an award already made.

    NIH published the policy as a companion pair: the formal Guide notice (NOT-OD-25-132) and an accompanying Extramural Nexus explainer, “Apply Responsibly,” released two weeks later on 31 July 2025. A supporting FAQ page clarifies edge cases such as how applications are counted and which activity codes are exempt.

    The Six-Application PI Limit, Explained

    The second, less-discussed half of the notice caps submissions. An individual PI or Program Director — including each PI listed on a Multiple PI application — may submit a maximum of six new, renewal, resubmission, or revision applications within a defined NIH Council Round. NIH has said the cap was prompted by evidence that a small number of investigators, aided by AI drafting tools, were submitting dozens of applications in a single round, in some cases reportedly more than 40, straining the peer-review system disproportionately relative to the scientific contribution.

    Council Round Applications counted toward the cap
    2026 Council Round Due dates 25 September 2025 through 7 May 2026
    2027 Council Round Due dates 25 May 2026 through 7 May 2027

    Two activity types are explicitly carved out: T-series training grants and R13 conference grants. The limit applies only to the PI/PD role — an investigator can still appear as a co-investigator, senior/key personnel, or a PI on a subaward across as many applications as the science supports, without those roles counting against their own six-application ceiling.

    • Counts toward the cap: new, renewal, resubmission, and revision applications where the individual is listed as PD/PI or Multiple PI.
    • Does not count toward the cap: T-activity-code training grants, R13 conference grants, and any application where the individual’s role is co-investigator or subaward PI rather than the lead PD/PI.
    • Counting trigger: institutional guidance (e.g., UCSF’s Office of Sponsored Research) indicates an application is counted toward the limit once it passes administrative review, prior to peer review — so late withdrawals after that point may still occupy a slot.

    PIs are responsible for tracking their own running total through eRA Commons; NIH does not promise to intercept a seventh application before submission, and rejection at that stage disrupts a funding cycle an institution cannot easily recover.

    AI Detection and Post-Award Consequences

    The most consequential detail for research offices is timing: NIH’s review for AI-substantiality is not confined to the submission window. The notice makes clear that detection can occur at the post-award stage — after funds have already been drawn down — and that a finding of substantial AI authorship at that point is treated with the same seriousness as post-hoc discovery of plagiarism or fabricated data.

    That reframes the risk calculus for institutions. A proposal that clears peer review and receives an award is not retroactively safe; institutional research integrity offices should treat AI-authorship disclosures and certifications as live documents that could be tested years into a project period, not a one-time submission checkbox.

    Compliance FAQs for Research Offices

    What is NOT-OD-25-132?

    NOT-OD-25-132 is an NIH Guide notice, “Supporting Fairness and Originality in NIH Research Applications,” issued 17 July 2025. It bars grant applications substantially developed by AI and, separately, caps most PIs at six NIH applications per year, effective for due dates from 25 September 2025.

    How many grants can one PI submit to NIH per year?

    A maximum of six new, renewal, resubmission, or revision applications per PD/PI or Multiple PI, counted per NIH Council Round rather than the calendar year strictly. T-series training grants and R13 conference grants are exempt from the count entirely.

    Does NIH ban all use of AI in grant writing?

    No. NIH restricts applications “substantially developed by AI” — content that is not the applicant’s original scientific thinking — rather than incidental editing or drafting assistance. The concern is plagiarism, fabricated citations, and misleading content, not tool use itself.

    What happens if AI use is detected after an award is made?

    NIH can pursue a research misconduct review, refer the case to the Office of Research Integrity, and take enforcement action including disallowance of costs or termination of the award — even after funding has already begun.

    Compliance Steps Before the Next Submission Cycle

    Research offices working toward the 2027 Council Round have a defined window to close process gaps. Practical steps that institutional sponsored-programs and research-integrity functions are adopting:

    • Add an explicit AI-originality certification to internal routing and sign-off forms, distinct from the existing research-misconduct and conflict-of-interest attestations.
    • Track each PI’s running six-application count centrally rather than relying on individual faculty to self-monitor eRA Commons, particularly across multi-department or multi-PI proposals.
    • Flag applications naming the same PI as PD/PI on more than six efforts early in the planning cycle, and confirm which, if any, qualify for the T-code or R13 exemption before assuming a conflict exists.
    • Brief investigators that post-award AI-authorship findings carry the same institutional exposure as post-award data-integrity findings — this is not solely a pre-submission compliance question.
    • Coordinate with departmental research administrators (the audience ARMA, NCURA, and EARMA serve) so that the cap is applied consistently across schools and centres within one institution, since NIH counts at the individual-PI level regardless of departmental structure.

    Implications for the Wider Research Ecosystem

    NOT-OD-25-132 sits inside a broader pattern: funders are converging on the idea that AI-assisted productivity gains in proposal writing are not neutral if they change who gets reviewed and how thoroughly. The six-application cap is, functionally, a peer-review capacity-protection measure as much as an integrity measure — NIH is acknowledging that AI tools changed submission economics faster than its review infrastructure could absorb.

    For research administration more broadly, the notice is also a signal that funder-side AI governance is arriving as binding operational policy, not aspirational guidance, and that institutions without a documented AI-in-proposals policy of their own are now exposed to funder-level enforcement they cannot pre-empt internally. Institutions that build durable certification and tracking workflows now — rather than treating this as a one-off notice to file away — will be better positioned as other US and international funders publish comparable AI-authorship and submission-volume rules over the next funding cycles.

  • AI in Grant Peer Review: How ERC, NIH, UKRI and NHMRC Draw the Line

    Four major funders have now published, or are actively revising, formal rules on AI in grant peer review, and the details differ enough that a reviewer moving between panels could unknowingly breach one funder’s terms while complying with another’s. In March 2026 the European Research Council (ERC) issued new guidelines on AI use in evaluation; the US National Institutes of Health (NIH) tightened its stance on AI-drafted applications from September 2025; UK Research and Innovation (UKRI) maintains a stricter blanket ban that peers expect to loosen; and Australia’s National Health and Medical Research Council (NHMRC) introduces a revised generative-AI policy from 28 April 2026. Research offices drafting or updating reviewer agreements need to track all four.

    How ERC, NIH, UKRI and NHMRC draw the line

    Each funder separates permitted “AI-assisted” support from prohibited “AI-generated” evaluation, but the exact boundary — and the effective date — varies.

    Funder Rule effective AI-assisted (permitted) AI-generated (prohibited)
    ERC 24 March 2026 Language polishing of a reviewer’s own report; general (non-proposal) information searches Summarising proposals, assessing scientific merit, drafting evaluations, uploading any proposal content to external AI systems
    NIH Applications submitted from 25 September 2025 Limited administrative tasks in application preparation Reviewers using generative AI to analyse applications or formulate critiques; applications “substantially developed by AI” are treated as non-original and not reviewed
    UKRI Current policy; Research Funding Policy Group review pending None yet formally sanctioned for reviewers — even AI-assisted grammar checks are currently disallowed Any generative AI use by reviewers or panellists in assessing applications
    NHMRC 28 April 2026 Generative AI to refine clarity or grammar of a reviewer’s own comments Using AI to evaluate, critique or score applications

    A fifth data point is worth noting: the US-based Foundation for Food & Agriculture Research (FFAR) went further still in November 2025, prohibiting reviewers from using AI tools in any capacity during peer review — including refinement of their own comments — on confidentiality grounds. That makes FFAR the strictest outlier against which UKRI’s current position, and NHMRC’s narrower allowance, can be benchmarked.

    • Confidentiality is the universal red line. Every policy reviewed prohibits uploading proposal text, applicant data or reviewer notes into public or third-party AI tools.
    • Non-delegation is the second constant. Scientific merit assessment must remain a human judgement in all four jurisdictions, regardless of how permissive the language-polishing allowance is.
    • UKRI is currently the most conservative of the four, with a sector-wide Research Funding Policy Group review expected to permit limited generative AI use in processing (not scoring) applications while keeping final decisions human-made.

    AI-assisted vs AI-generated: common questions

    Research offices repeatedly ask the same handful of questions when briefing reviewers. The answers below are grounded in the funder documents referenced above.

    What is the difference between AI-assisted and AI-generated peer review?

    AI-assisted review means a human reviewer uses a tool only for mechanical tasks — grammar, clarity, formatting of their own text — while retaining full intellectual authorship of the assessment. AI-generated review means the AI performs part of the evaluative task itself, such as summarising a proposal, scoring merit, or drafting critique content, which every funder surveyed here prohibits.

    Has NIH banned AI in grant peer review?

    Yes. NIH prohibits scientific peer reviewers from using generative AI tools to analyse applications or formulate critiques, a position it has held since June 2023. From 25 September 2025, NIH also treats applications substantially developed by AI as non-original, removing them from review rather than scoring them on merit.

    Can UKRI reviewers use AI to check grammar in their assessments?

    Not currently. UKRI’s existing policy forbids reviewers and panellists from using generative AI for any part of assessment, including language or grammar correction — a stricter line than ERC or NHMRC. A sector-wide funder policy group is expected to revisit this, but any change would still require human-made final decisions.

    When does the NHMRC generative AI policy take effect?

    NHMRC’s revised Policy on Use of Generative Artificial Intelligence in Grant Review takes effect from 28 April 2026. It permits peer reviewers to use generative AI to refine the clarity or grammar of their own comments, but explicitly prohibits using AI to evaluate, critique or score applications.

    Practical reviewer-agreement language for research offices

    Research offices administering panels — whether for an internal seed-fund competition, a co-funded international call, or as a delegated peer-review manager for an external funder — need reviewer agreements that anticipate divergence between funder rules. Three drafting principles reduce risk:

    • Name the prohibited actions explicitly, not just the tool category. A clause banning “AI tools” is weaker than one banning “uploading proposal content, applicant identifiers, or draft scores to any AI system, whether or not the funder’s own policy names that system.”
    • State the confidentiality obligation independently of the AI-use clause. General-purpose AI (GPAI) providers regulated under the EU AI Act’s GPAI obligations, in force since August 2025, may process submitted inputs for model improvement unless expressly excluded, so agreements should require reviewers to confirm no proposal content has been shared with any third-party system, GPAI-regulated or not.
    • Require disclosure, not just prohibition. A short attestation line — “I have not used generative AI to draft, summarise or score any part of this review, and any AI assistance used was limited to language editing of my own original text” — gives research integrity offices an auditable record if a dispute arises.

    Where a funder (such as NHMRC from April 2026) permits limited AI-assisted editing, research offices should still require reviewers to disclose which tool was used and confirm no proposal content was entered into it. This keeps institutional practice defensible even where funder rules differ from one call to the next.

    Implications and outlook

    For institutions running multi-funder portfolios, the practical challenge is less about any single funder’s rule and more about reviewer confusion across simultaneous panels. A reviewer serving both an ERC panel and a UKRI-funded call in the same month operates under materially different AI permissions for the same underlying task. Research offices should treat funder AI policies as living documents — ERC’s and NHMRC’s 2026 updates both followed roughly a year or more after their organisations’ initial public positions on AI, suggesting further revision is likely as reviewer behaviour and AI capability both evolve.

    The direction of travel across all four funders is convergence on two non-negotiables — confidentiality of proposal content and non-delegation of scientific judgement — even as the permitted margin for administrative AI assistance slowly widens. Research offices that build reviewer agreements around those two constants, rather than around any single funder’s current wording, will need fewer rewrites as UKRI’s pending policy shift and any subsequent NIH or ERC revisions land through 2026 and beyond.

    For related terminology used across funder and publisher AI-governance documents, see the CASRAI research dictionary, and for broader institutional process guidance visit the research administration resource hub.