{"resourceId":"umd-athletics-genai-data-governance","versions":[{"version":"external-c6d5a6638f4aeb383d8d932ef7a05dff4c06517c2e1df099e5641f696510a7dc","resource":{"id":"umd-athletics-genai-data-governance","title":"Maryland Athletics publishes GenAI Data Governance with four levels and Golden Rules","organization":"University of Maryland Athletics IT","sector":"Higher education athletics / IT governance","geography":"United States (College Park, Maryland)","publishedAt":"undated page retrieved 2026-09-28","publicationDate":null,"eventDate":null,"sourceName":"Maryland Athletics IT","sourceLabel":"Institutional athletics IT guidance","sourceUrl":"https://athleticsit.umd.edu/ai-initiative/ai-data-governance","evidenceClass":"standards-guidance","outcomeClass":"cautionary","topics":["data-security","governance-procurement","knowledge-work","operating-model"],"finding":"Maryland Athletics IT publishes a GenAI Data Governance guide organizing athletics data into four levels (Level 1 public through Level 4 restricted—no GenAI) so staff can decide what is safe to share, what stays internal, and what needs approval. The sheet includes do's/don'ts and UMD Data & AI Golden Rules: classify first, minimize what you share, use the right approved tool, verify the output, and avoid bulk exports, plus escalation to Athletics IT for a fast data check. No adoption metrics or audit results are reported. Why now: practical staff controls as peer departments adopt GenAI content/ops tools.","sledRelevance":"Operational GenAI control framework inside a Power conference athletic department—directly addresses student-athlete and institutional data risk in everyday AI use.","evidence":"Athletics IT page describing four-level classification, Golden Rules list, and Athletics IT escalation path; companion AI Initiative site confirms DIT partnership context.","architectureImplications":"Policy overlay on approved GenAI tools rather than a new model stack; requires mapping athletics systems to classification levels and routing restricted data away from public LLMs.","governanceImplications":"Makes classify-first and human verification explicit departmental norms; pairs with campus DIT authority and unit-level accountability.","securityPrivacyImplications":"Level 4 ban on GenAI for restricted data is the key control; minimize-share and no-bulk-export rules target prompt leakage of student-athlete and institutional data.","caveats":"Undated page; detailed sheet is image-hosted; enforcement and audit outcomes not published.","streamIds":["college-athletics"],"roles":{"sales":"Athletic department staff adopting Copilot/ChatGPT need clear rules for what institutional and student-athlete data may enter GenAI tools. Discovery should map current shadow AI use, which systems hold Level 3/4 data, and how Athletics IT escalations work. Bound value is a practical four-level classification guide with Golden Rules—not a claim that Maryland eliminated AI risk. Position governance enablement alongside approved tools. Unsupported claims include comprehensive FERPA/HIPAA compliance certification from this one-pager alone or zero residual leakage risk.","engineering":"Fit is a staff-facing GenAI data-classification control framework (Levels 1–4) with approved-tool routing and Athletics IT escalation—not a model deployment. Integration means aligning athletic systems' data labels with campus DIT classes and blocking Level 4 (restricted) from GenAI. Prerequisites: inventory of athletics data stores and approved enterprise AI tools. Constraints: detailed rules live partly in a downloadable sheet/image; enforcement depends on staff behavior. Security centers on minimize-share, no bulk export, and verify-output norms. Proof of value: escalation ticket volume, policy quiz pass rates, and sampled prompt audits.","delivery":"Delivery work is publishing and training on the GenAI Data Governance sheet, embedding Golden Rules into AI literacy sessions, and operating the Athletics IT data-check escalation path. Dependencies include campus DIT classifications and approved tool list. Ownership: Athletics IT with unit managers accountable for staff practice. Skills adoption: classify-first habit before any paste into GenAI. Governance checkpoints: quarterly sheet review, spot audits of high-risk units (medicine, compliance). Acceptance criteria: staff can state Level 4 ban; escalation path responds within agreed SLA. Risks: image-only details going stale and staff bypassing approved tools."},"roleDetails":{"sales":{"customerProblem":"Athletics staff lack a quick classification guide for what UMD data is safe to put into GenAI tools as AI use spreads.","stakeholders":["Athletics IT","Sports medicine / compliance","Coaches and communications staff","Campus Division of IT"],"discoveryQuestions":["Which athletics data stores contain Level 3/4 restricted information?","What enterprise GenAI tools are already approved?","How do staff escalate uncertain classification today?"],"valueHypothesis":"A four-level GenAI data guide with Golden Rules can reduce unsafe prompts if paired with approved tools and an Athletics IT check path.","potentialEngagement":"Governance enablement workshop plus 90-day prompt-audit pilot in communications and compliance units.","unsupportedClaims":["Complete elimination of data leakage","Legal certification of FERPA/HIPAA from the one-pager alone","Measured productivity gains from governance alone"]},"engineering":{"fit":"Policy/control overlay classifying athletics data Levels 1–4 for GenAI use, with approved-tool routing and escalation.","architectureIntegration":"Map athletics systems to classification levels; route work to approved GenAI; ticket path to Athletics IT for uncertain cases.","prerequisites":["Campus data classification scheme","Approved GenAI tool list","Staff-accessible governance sheet"],"constraints":["Enforcement is largely procedural/staff-driven","Detailed do/don't content is image-hosted","No published audit metrics on the page"],"security":"Level 4 restricted = no GenAI; minimize sharing; avoid bulk exports; verify outputs before use.","proofOfValue":["Staff classification quiz pass rate","Athletics IT escalation volume and time-to-answer","Sampled prompt audits in high-risk units"]},"delivery":{"work":"Publish/update governance sheet, train staff on Golden Rules, operate escalation inbox, refresh with DIT changes.","dependencies":["Campus DIT policy updates","Approved tool catalog","Unit manager reinforcement"],"ownership":"Athletics IT owns the guide and escalation; unit leaders own day-to-day compliance.","skillsAdoption":"Train classify-first, minimize-share, verify-output habits before any GenAI paste of athletics data.","governanceCheckpoints":["Quarterly sheet accuracy review","Spot audits in sports medicine and compliance","Annual alignment with campus AI policy"],"acceptanceCriteria":["Staff can identify Level 4 ban examples","Escalation path documented and monitored","Governance linked from AI training materials"],"risks":["Staff using personal ChatGPT on restricted data","Stale image sheet after policy changes","Over-blocking that drives shadow AI"]}},"retrievedAt":"2026-09-29T03:23:14Z","enrichedAt":"2026-09-29T03:25:30Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Gives non-technical athletics staff a short classification habit, but requires literacy training so rules are usable under deadline pressure.","procurementImplications":"Prefer enterprise-approved GenAI tools that support data-classification commitments; contract for no-training and retention controls aligned to Level rules.","operatingModelImplications":"Embeds Athletics IT as GenAI traffic cop via escalation path while expecting unit staff to classify before paste.","sourceVerification":{"openedUrl":"https://athleticsit.umd.edu/ai-initiative/ai-data-governance","referenceExcerpt":"classify first, minimize what you share, use the right approved tool, verify the output, and avoid bulk exports","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}