{"resourceId":"educause-ace-ai-procurement-interviews-2025","versions":[{"version":"external-d4c7de2a14af9100d84d954ab8aa9dab90a937ac2018989c18d925bafbfccc74","resource":{"id":"educause-ace-ai-procurement-interviews-2025","title":"Campus procurement interviews highlight hidden AI features, opaque terms and support capacity","organization":"EDUCAUSE and American Council on Education","sector":"Higher education procurement and IT governance","geography":"Primarily United States; includes varied institution types","publishedAt":"March 11, 2025","publicationDate":"2025-03-11","eventDate":null,"sourceName":"EDUCAUSE Review","sourceLabel":"Association interview synthesis and guidance","sourceUrl":"https://er.educause.edu/articles/2025/3/ai-procurement-in-higher-education-benefits-and-risks-of-emerging-tools","evidenceClass":"public-sector-association","outcomeClass":"cautionary","topics":["governance-procurement","data-security","accessibility-workforce","operating-model"],"finding":"The interview synthesis treats AI procurement as an ongoing review of products, embedded features and third-party use, with cost and transparency constraints.","sledRelevance":"Directly relevant to campus purchasing and small-team implementation capacity; specific state purchasing rules need local review.","evidence":"ACE and EDUCAUSE interviewed 12 CIOs, IT leaders and procurement professionals in October–November 2024. Their qualitative findings identify disclosure, pricing and capacity problems. The article also cites a separate survey; its percentages are not reused here. The opening Middlevale vignette is illustrative, not a real deployment.","architectureImplications":"Interpretation: inventory AI data paths across purchased and embedded services before approving integrations.","governanceImplications":"Interpretation: trigger reassessment when a vendor adds material capabilities, not only at contract signature.","securityPrivacyImplications":"Interpretation: obtain explicit answers about training reuse, retention, subprocessors and institutional data separation.","caveats":"Small qualitative interview sample; no causal estimate of procurement benefits, representative prevalence or current vendor pricing. Guidance is older than this edition and is not a statement of current legal requirements.","streamIds":["campus-operations"],"roles":{"sales":"Interpretation: ask procurement, counsel, the CISO and the business sponsor where approval stalls and what evidence they need to proceed. Inventory an existing product's newly added AI feature as a bounded engagement before considering an enterprise rollout. The value hypothesis is clearer decisions and fewer surprise obligations, not guaranteed savings. Ask whether staff capacity, recurring inference charges and exit effort are funded. The interview findings justify questions about disclosure and smaller-campus constraints; they do not establish that any particular vendor is unsuitable or that a sale is available.","engineering":"Interpretation: turn the vendor questionnaire into an evidence map linking each claim to a configuration, contract term or reproducible test. Require a sandbox and representative non-sensitive inputs. Trace prompts, retrieved data, logs and outputs across service boundaries, including embedded assistants and agents. Test role separation and feature-disable behavior rather than accepting an architecture diagram as assurance. Proposed validation should produce an explicit list of verified, unverified and inapplicable controls plus an estimate of integration effort. Deployment choice should follow data and service requirements, not an assumption that cloud or local hosting is inherently adequate.","delivery":"Interpretation: make the purchasing lead accountable for contract checkpoints and name an IT service owner for ongoing operation. Coordinate security, privacy, accessibility and business-user reviews before onboarding. Create a release-change register, renewal calendar and tested exit procedure; train staff to route unreviewed features to the owner. Proposed acceptance requires documented owners for material obligations, an approved support budget and successful export or deletion testing within the agreed scope. Track review cycle time and exceptions to improve the process. Risks include review overload, hidden feature activation and unfunded support after initial purchase."},"retrievedAt":"2026-09-09T03:01:11Z","enrichedAt":"2026-09-09T03:04:27Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Interpretation: include accessible task testing and funded support time in product evaluation.","procurementImplications":"Interpretation: negotiate testable obligations, change notification and feasible exit provisions using local counsel.","operatingModelImplications":"Interpretation: align purchasing, IT security and business ownership in one lifecycle review.","updateExplanation":"New URL across all archive pages. Included as foundational scrutiny of procurement implementation, which previous editions did not examine at this contract-detail level; not daily news.","sourceVerification":{"openedUrl":"https://er.educause.edu/articles/2025/3/ai-procurement-in-higher-education-benefits-and-risks-of-emerging-tools","referenceExcerpt":"Pricing and contract terms are also opaque.","promptVersion":"sled-research-v3.1","model":null,"basis":"agent-reported inspection"}}}]}