{"resourceId":"cornell-policy-1-8-responsible-ai-research-20261102","versions":[{"version":"external-789b4312e72fb4b11f52795054760bd6e05fbfabd7ac2995031f355a579a9ac4","resource":{"id":"cornell-policy-1-8-responsible-ai-research-20261102","title":"University policy on AI in research published (Cornell University Policy 1.8)","organization":"Cornell University (Cornell Data Services; University Policy Office)","sector":"University research governance (institutional AI-in-research policy)","geography":"United States (New York)","publishedAt":"2 November 2026 (Cornell Data Services news)","publicationDate":"2026-11-02","eventDate":null,"sourceName":"Cornell Data Services","sourceLabel":"Cornell Data Services, \"University policy on AI in research published,\" 2 Nov 2026","sourceUrl":"https://data.research.cornell.edu/2026/11/02/university-policy-on-ai-in-research-published/","evidenceClass":"standards-guidance","outcomeClass":"emerging","topics":["governance-procurement","data-security","knowledge-work"],"finding":"Cornell published University Policy 1.8, Responsible and Ethical Use of AI in Research and Scholarship, to give researchers principles for AI use and to identify their responsibilities for rigor and integrity. Section 3.2 on data directs researchers to use institution-approved tools when available, avoid uploading private, export-controlled, sensitive or proprietary data to general AI platforms, verify tools' data-sharing practices, share high- or moderate-risk institutional data only with internally reviewed tools, minimize personal or confidential data, and take reasonable security measures.","sledRelevance":"Directly relevant to research universities drafting AI-in-research policy: a peer institution has moved from guidance to formal policy, tying data-risk tiers to tool review. It is a policy, not evidence of compliance or outcomes.","evidence":"Institutional policy announcement summarizing one section; the full policy text was not opened. No implementation data, enforcement mechanism, or approved-tool list in the post.","architectureImplications":"Implies an institutional approved-tool catalogue and internal review process for AI tools handling moderate/high-risk data.","governanceImplications":"Formal policy assigns responsibility to researchers; institutions need review processes and data classification to make it workable.","securityPrivacyImplications":"Data-tier rules restrict sensitive, export-controlled and proprietary data to reviewed tools; minimization and security measures required.","caveats":"Announcement summary only; full policy and other sections not inspected; no evidence of adoption or enforcement.","streamIds":["research"],"roles":{"sales":"Governance framing: a peer research university has formalized AI-in-research rules that tie data risk tiers to approved, reviewed tools. That supports bounded conversations with research offices about tool review processes and approved-tool catalogues. Interpretation: a policy announcement cannot support claims that it reduces incidents or that any particular tool qualifies.","engineering":"Fit is institutions needing to route moderate- and high-risk research data only to reviewed AI tools. Integration involves data classification, an approved-tool catalogue, contract review of data-sharing terms and identity-based access. Prerequisites are existing data-risk tiers and a review process. Constraints: the full policy was not inspected. A proof of value would map current researcher AI tools against data tiers and review status.","delivery":"Delivery work is operationalizing an AI-in-research policy: tool review workflow, approved-tool list, researcher guidance and training, and exception handling. Ownership sits with the research office and IT security or privacy office. Governance checkpoints include tool reviews and periodic policy updates. Acceptance criteria are proposals such as published approved-tool list and review turnaround; risks include shadow AI use if approved tools lag needs."},"roleDetails":{"sales":{"customerProblem":"Researchers upload sensitive or controlled data to general AI tools without clear institutional rules or reviewed alternatives.","stakeholders":["Vice President for Research","Research integrity officer","CISO / privacy officer","Export control office","Faculty senate"],"discoveryQuestions":["Does your institution have an AI-in-research policy or only guidance?","Which AI tools are reviewed for moderate- or high-risk data?","How are export-controlled data protected from general AI platforms?","How quickly can new tools be reviewed?"],"valueHypothesis":"A policy tied to data tiers and reviewed tools may give researchers clearer paths to compliant AI use.","potentialEngagement":"A gap assessment comparing an institution's AI-in-research guidance and tool review against a peer policy such as Cornell 1.8.","unsupportedClaims":["Reduced incidents or violations","Compliance of any specific tool","Completeness of the policy beyond Section 3.2"]},"engineering":{"fit":"Fits research universities with data classification schemes.","architectureIntegration":"Approved-tool catalogue, tool review workflow, data classification, identity and access controls.","prerequisites":["Data-risk tiers","Tool review process","Contract terms on data sharing"],"constraints":["Full policy not inspected","No enforcement data","Tool landscape changes quickly"],"security":"Restrict sensitive and export-controlled data to reviewed tools; enforce minimization.","proofOfValue":["Inventory tools in use against data tiers","Time a tool review end to end","Sample researchers on awareness"]},"delivery":{"work":"Stand up tool review and approved-tool list; Publish researcher guidance and training","dependencies":["Research office","IT security and privacy","Procurement and legal"],"ownership":"Research office, jointly with IT security/privacy.","skillsAdoption":"Researchers and staff need training on data tiers and approved tools.","governanceCheckpoints":["Tool review sign-off","Annual policy review","Exception process"],"acceptanceCriteria":["Proposed: approved-tool list published and maintained","Proposed: tool review turnaround measured","Proposed: training completion tracked"],"risks":["Shadow AI use","Review backlog","Policy drift as tools change"]}},"retrievedAt":"2026-11-04T22:01:18-06:00","enrichedAt":"2026-11-04T22:01:19-06:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Researchers need data-tier and AI literacy training to follow the policy.","procurementImplications":"Approved tools imply procurement and contract review of data-sharing terms.","operatingModelImplications":"Research office and IT jointly own policy operation.","sourceVerification":{"openedUrl":"https://data.research.cornell.edu/2026/11/02/university-policy-on-ai-in-research-published/","referenceExcerpt":"avoid uploading private, export-controlled, sensitive, or proprietary data to general AI platforms","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}