{"resourceId":"gatech-admin-ai-employment-policy-202606","versions":[{"version":"external-a01afccbc88eff0da41d5194d7de5f4475ff07ebc722a13fc863222957c61bb9","resource":{"id":"gatech-admin-ai-employment-policy-202606","title":"Georgia Tech drafts admin/employment AI policy with registers, human review, and suspension powers","organization":"Georgia Institute of Technology","sector":"Higher education administration and employment","geography":"United States (Georgia)","publishedAt":"Policy Steering Committee Approval June 2026 (Effective Date TBD)","publicationDate":"2026-06-01","eventDate":null,"sourceName":"Georgia Tech Policies","sourceLabel":"Georgia Tech AI in Administrative and Employment Contexts Policy PDF","sourceUrl":"https://policies.gatech.edu/sites/default/files/2026-07/Artificial%20Intelligence%20%28AI%29%20in%20Administrative%20and%20Employment%20Contexts%20Policy.pdf","evidenceClass":"standards-guidance","outcomeClass":"emerging","topics":["governance-procurement","data-security","accessibility-workforce","operating-model"],"finding":"June 2026-approved (effective date TBD) Georgia Tech policy for administrative and employment AI creates an Institute AI Governance Program and Committee, routes tools to Institute or Local AI Registers after risk-based review, restricts Protected Data and PII to verified tools, prohibits automating employment, grievance, or disciplinary decisions without appropriate human review, and authorizes suspension of tools causing severe operational, privacy, or security consequences, with Internal Audit compliance authority.","sledRelevance":"Control blueprint for campus-operations AI: registers, employment-action limits, privacy guardrails, and audit/suspension powers complementary to labor demands for human authority.","evidence":"Primary institutional policy PDF. Steering Committee approval June 2026; Effective Date still TBD, so operational enforceability may not yet be active.","architectureImplications":"Interpretation: inventory-first architecture—no tool use until register approval; local limited-use registers for constrained tools; intelligent automations may need lifecycle logs and go/no-go approvals.","governanceImplications":"Interpretation: cross-functional committee plus Internal Audit creates enforceable governance beyond guidance memos; employment automation without human review is explicitly out of bounds.","securityPrivacyImplications":"Protected Data and PII only in tools with verified security controls; CISO suspension authority for severe consequences; prompt-injection and hallucination safeguards listed as principles.","caveats":"Effective date TBD; outcomes of compliance unknown; USG-specific framing; retrieved PDF includes a DRAFT marker in a summary header area.","streamIds":["campus-operations"],"roles":{"sales":"Campuses asking what good administrative AI governance looks like can use Georgia Tech’s register-plus-human-review model as a requirements checklist while noting the effective date is still TBD. The June 2026-approved policy creates an Institute AI Governance Program, Institute and Local AI Registers, Protected Data gates, a ban on automating employment actions without human review, and suspension powers with Internal Audit authority. Position offerings that support inventory, approval workflows, and employment-decision human review. Do not claim the policy is already fully in force or that registers alone stop shadow AI.","engineering":"Implement AI registers, risk routing, Protected Data controls, and kill-switch or suspension hooks tied to cybersecurity assessments and procurement intake. No production administrative tool should run until it is evaluated and listed. Prerequisites include a governance officer, data classification, an intake form, and a tested suspension runbook. Constraints: effective date TBD, University System of Georgia policy-stack dependencies, and local-register consistency risk. Security requires verified controls before Protected Data use, CISO suspension for severe incidents, and prompt-injection awareness listed among guiding principles.","delivery":"Stand up register processes, migrate known tools, train managers that employment, grievance, or disciplinary automation needs human review, and enable Internal Audit sampling. Ownership is the AI Governance Program under the EVP for Administration and Finance, with OIT, HR, procurement, and Internal Audit as partners. Acceptance: sampled units show no unregistered production admin AI, documented human review for employment-impacting tools, and a tested suspension path. Risks include policy lag while the effective date remains TBD, local-register fragmentation, and shadow tools outside intake."},"roleDetails":{"sales":{"customerProblem":"Administrative AI spreads faster than inventories, employment safeguards, and auditability.","stakeholders":["EVP Administration/Finance","AI Governance Officer","CISO","HR","Internal Audit","Procurement"],"discoveryQuestions":["Is there an AI register today?","Who can suspend a tool?","Are employment decisions human-reviewed?","How is Protected Data gated?"],"valueHypothesis":"Register-based approval with employment human-review rules reduces unmanaged admin AI risk; Georgia Tech documents the control pattern pending effective date.","potentialEngagement":"Policy-to-operating-procedure implementation for AI registers and employment-use controls.","unsupportedClaims":["Policy already fully in force","Registers alone prevent shadow AI","Guaranteed bias-free employment tools"]},"engineering":{"fit":"Enterprise admin and employment AI governance controls for a research university.","architectureIntegration":"Central and local AI registers; risk-based approval; integration with cybersecurity assessments and procurement intake.","prerequisites":["Governance committee charter","Data classification","Tool inventory intake form","Suspension runbook"],"constraints":["Effective Date TBD","USG policy stack dependencies","Local register consistency risk"],"security":"Verified controls before Protected Data use; CISO suspension for severe incidents; prompt-injection awareness.","proofOfValue":["Percent of admin AI tools on register","Time-to-approval metrics","Tabletop of suspension authority"]},"delivery":{"work":"Publish procedures, migrate known tools onto registers, train managers on employment-use bans without human review, enable audit sampling.","dependencies":["OIT/Cybersecurity","HR","Procurement","Internal Audit"],"ownership":"AI Governance Program under EVP Administration and Finance.","skillsAdoption":"Mandatory compliance training and self-service literacy resources as policy promotes.","governanceCheckpoints":["Register listing before production use","High-risk automation go/no-go","Periodic Internal Audit review"],"acceptanceCriteria":["No unregistered production admin AI in sampled units","Documented human review for employment-impacting AI","Tested suspension path"],"risks":["Policy lag because effective date is TBD","Local register fragmentation","Shadow tools outside intake"]}},"retrievedAt":"2026-09-15T18:48:51Z","enrichedAt":"2026-09-15T18:57:40Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Accessibility is a listed ethical principle; employment, grievance, and discipline automation limits protect workers from opaque AI decisions. Training expectations create workload for staff.","procurementImplications":"AI tools must pass register approval; procurement intake should block purchases that skip governance review.","operatingModelImplications":"Central governance program with local registers balances enablement and control for administrative AI.","sourceVerification":{"openedUrl":"https://policies.gatech.edu/sites/default/files/2026-07/Artificial%20Intelligence%20%28AI%29%20in%20Administrative%20and%20Employment%20Contexts%20Policy.pdf","referenceExcerpt":"privacy risks, or automated employment actions without human review.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}