{"resourceId":"amrc-using-ai-in-research-funding-guidance-funders-human-decisions-20261102","versions":[{"version":"external-945144ec1081dc20f1d56175ae72ceb810fad500579396678ca43feaff1c7b79","resource":{"id":"amrc-using-ai-in-research-funding-guidance-funders-human-decisions-20261102","title":"Using AI in research funding: guidance for funders and reviewers (Association of Medical Research Charities)","organization":"Association of Medical Research Charities (AMRC)","sector":"Research funding (medical research charities)","geography":"United Kingdom (applicable to funders and reviewers more widely)","publishedAt":"2 November 2026 (updated 4 November 2026)","publicationDate":"2026-11-02","eventDate":null,"sourceName":"AMRC","sourceLabel":"Association of Medical Research Charities, \"Using AI in research funding,\" published 2 Nov 2026, updated 4 Nov 2026","sourceUrl":"https://www.amrc.org.uk/using-ai-in-research-funding","evidenceClass":"public-sector-association","outcomeClass":"cautionary","topics":["governance-procurement","knowledge-work","data-security"],"finding":"The UK association for medical research charities says generative AI could help funders and reviewers with some tasks but must not replace human judgment, and states that it must not be used to make research funding decisions. Listed possible funder uses include drafting application-form structures and email replies, an applicant question chatbot, identifying potential expert reviewers, drafting meeting minutes and applicant feedback, lay summaries and translation. Guidance says to consider tool appropriateness, protection of personal, confidential and scientific information and intellectual property, whether to set funder staff policies, and how to tell applicants, reviewers and supporters about AI use. It recommends confidential in-house or enterprise data-protected tools assessed for security and says confidential applications must not be entered in publicly available tools. Reviewers should be told not to enter applications into generative AI tools, though some funders may allow limited uses such as language refinement under stated conditions. Included now because it was published 2 Nov (updated 4 Nov), was not archived, and is funder-side guidance that complements last week's archived Springer review of funder AI policies.","sledRelevance":"Relevant to university sponsored-programs and research offices that advise applicants and reviewers, and to research funders: it states what a sector body expects for confidentiality, human accountability and disclosure, which shapes tool choices and internal policy.","evidence":"Sector-association web guidance (not empirical research). It sets expectations and considerations; it reports no measured outcomes, sample or evaluation method. Author and dates are stated on the page.","architectureImplications":"Where AI supports funding workflows, use confidential in-house or enterprise data-protected tools assessed for security, review how each tool retains and could reuse submitted information, and keep tools separate from systems that hold unpublished proposals.","governanceImplications":"People stay responsible for decisions; AI use needs appropriate trialing and safeguards, funder-defined rules for staff and reviewers, and transparent communication to applicants and reviewers. Reviewer entry of applications into generative AI tools should be prohibited, with any limited allowed uses specified.","securityPrivacyImplications":"Confidential funding applications, personal data and scientific ideas must not go into publicly available tools; intellectual-property and confidentiality protections depend on the tool's data handling terms.","caveats":"Guidance, not evidence of effectiveness; UK medical research charity context; not binding on members or other funders; no measured results; the page says 'Published: November 2026' in its summary and 2 November in the footer, with an update on 4 November.","streamIds":["research"],"roles":{"sales":"Framing point: a UK sector body states plainly that generative AI must not make research funding decisions and that confidential applications belong only in assessed in-house or enterprise data-protected tools. Use it with sponsored-programs, research-administration and funder-facing teams building AI policy. Position a bounded policy-and-tooling assessment covering approved tools, reviewer rules and disclosure wording, not a promise of efficiency gains; the guidance reports no measured results and applies to charity funders, so confirm which funder policies your institution must follow.","engineering":"Framing point: the technical requirement is data protection, not model capability. Tools touching grant applications must be confidential, security-assessed and clear about retention and reuse, so route them through enterprise tenancy or in-house hosting with logging and no training on submitted content. Public chat tools stay blocked for applications. Missing from the guidance: detailed controls, audit design and test criteria. Validate with data-flow review and retention checks before enabling any drafting or chatbot use.","delivery":"Framing point: the work is policy, communication and training. Draft rules for staff and reviewers, define the limited uses allowed (for example language refinement), publish disclosure wording for applicants and brief reviewers on what must not be entered into AI tools. Assign a policy owner in research administration and review decisions when funders change rules. The page gives no schedule, cost or adoption data, so treat proposed milestones as planning assumptions."},"roleDetails":{"sales":{"customerProblem":"Funders and research offices lack clear rules on where AI may be used with confidential applications and review processes.","stakeholders":["Vice president for research","Sponsored programs director","Research compliance and integrity officers","Funder program and grants managers"],"discoveryQuestions":["Which funder AI rules apply to our applicants and reviewers?","Which AI tools currently touch proposals?","Who approves tools for confidential material?","How do we communicate AI use to applicants?"],"valueHypothesis":"A clear tool-and-policy framework may reduce confidentiality risk and confusion for applicants and reviewers; the guidance does not measure benefits.","potentialEngagement":"A short policy and tooling assessment mapping current AI use in proposal support and review, funder rules and approved data-protected tools.","unsupportedClaims":["Efficiency or cost savings","Compliance with any specific funder's rules","That limited reviewer uses are permitted everywhere"]},"engineering":{"fit":"Relevant wherever proposals or reviewer material may reach AI tools; it is policy guidance rather than a technical specification.","architectureIntegration":"Route any AI assistance for grants through enterprise or in-house tools with confirmed non-reuse of submitted data; block public tools for applications through network or endpoint controls.","prerequisites":["Inventory of AI tools in use","Data-handling terms for each tool","Data classification for proposals"],"constraints":["Guidance gives no technical controls","Funder rules vary","Reviewer limits depend on each funder"],"security":"Confidential applications, personal data and unpublished ideas must stay out of public tools; evaluate retention and reuse terms.","proofOfValue":["Data-flow review of each AI tool touching proposals","Test that public tools are blocked for application content","Confirm no training or reuse in vendor terms"]},"delivery":{"work":"Draft staff and reviewer rules, approve tools, publish disclosure language and train applicants, staff and reviewers.","dependencies":["Funder policy review","Security assessment of tools","Leadership approval of policy"],"ownership":"Research administration or sponsored programs owns policy; IT security owns tool assessment.","skillsAdoption":"Staff and reviewers need short training on what may and may not be entered into AI tools and how to disclose use.","governanceCheckpoints":["Policy approval before tool enablement","Periodic review against funder rule changes","Incident process for confidential data entered into public tools"],"acceptanceCriteria":["Proposed: approved-tool list published","Proposed: reviewer guidance issued before the next review cycle","Proposed: disclosure wording available to applicants"],"risks":["Applicants or reviewers using public tools","Conflicting funder rules","Policy lagging tool changes"]}},"retrievedAt":"2026-11-09T21:56:09-06:00","enrichedAt":"2026-11-09T21:56:09-06:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Limited language-refinement uses may help non-native English speakers, which funders can allow under stated conditions; guidance asks funders to communicate expectations to reviewers.","procurementImplications":"Specify confidential or enterprise data-protected tools, assess security and safety, and review retention and reuse before purchase for any grant-support use.","operatingModelImplications":"Keeps humans accountable for funding decisions and places tool assessment, reviewer rules and applicant communication with the funder or research office.","peerReview":"unknown","sourceVerification":{"openedUrl":"https://www.amrc.org.uk/using-ai-in-research-funding","referenceExcerpt":"generative AI must not be used to make research funding decisions","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}