CSU GenAI Advisory Committee 2026 report (web)
Citation
California State University (CSU), Generative AI Advisory Committee. “Generative AI Advisory Committee – 2026 Report.” June 2026 Final, 22 pages, PDF. https://genai.calstate.edu/sites/default/files/2026-09/GenAI%20Advisory%20Committee%20%E2%80%93%202026%20Report%20FINAL.pdf. Read 2026-11-09.
Bottom line
- The CSU generative artificial intelligence (GenAI) committee gives seven recommendations for the next phase of AI use in the CSU system.
- It recommends that the CSU renew ChatGPT EDU now and move to a “multi-LLM or gateway-based model” later. It also recommends one systemwide AI governance, risk and procurement framework.
Background
The CSU Generative AI Advisory Committee has three subcommittees: Teaching, Learning & Scholarship; Privacy, Security & Procurement; and Enhanced Productivity Applications. The report gives the committee’s recommendations, a three-phase roadmap, the membership rosters and each subcommittee’s own report. The Chancellor’s Office Chief Information Officer and an Associate Vice Chancellor for Academic Affairs co-chair the committee. Appendix C describes the CSU AI Insights Framework, built with the strategic partner Milestone Technologies.
Key claims
- “Across these groups, there is strong alignment that the CSU must now transition from decentralized experimentation to a coordinated, systemwide strategy.” (Executive Summary, p. 1)
- “The CSU should renew ChatGPT EDU in the near term to ensure continuity in teaching, learning, and administrative use, while simultaneously developing a longer-term, multi-vendor AI strategy.” (Recommendation 5, p. 5)
- “This recommendation includes developing sandbox environments to test alternative tools, establishing evaluation criteria for AI technologies, and creating a transition path toward a multi-LLM or gateway-based model.” (Recommendation 5, p. 5)
- “The Privacy, Security & Procurement subcommittee identified significant gaps in current practices, including inconsistent contract language, limited vendor disclosure of AI capabilities, and insufficient safeguards around vendor use of CSU data.” (Recommendation 4, p. 5)
- “Existing contracts do not consistently restrict vendor model training on CSU data.” (Privacy, Security & Procurement Subcommittee, Year 1 accomplishments, p. 17)
- “Develop a sandbox environment to test alternative, open-source, and discipline-specific models” (Teaching, Learning and Scholarship Subcommittee, recommendation 4, p. 13)
- “Phase 3: Optimization and Scale (24+ months) • Transition toward multi-LLM or gateway architecture” (Implementation Roadmap, p. 7)
Evidence and method
A committee report with recommendations. It gives no budget, cost per user or usage figures. Its evidence is the subcommittees’ review of CSU policy, contracts and campus practice in 2025-26.
Limits
The report does not name a hardware or cloud platform for the sandbox or the gateway. It does not give the cost of the ChatGPT EDU renewal. See CSU GenAI FAQ (web) for the per-user price and the license count.