UC Santa Barbara is giving its campus community access to AI from multiple providers, rather than making one vendor the university’s sole option. In an October 12 announcement, UCSB described recent campuswide rollouts of Google Gemini and its AI Commons platform, pairing hands-on experimentation with campus-approved privacy boundaries and training in responsible use.
Different research, different models
AI Commons is built around multiple vendors and models. It provides access to commercial systems alongside open-source and open-weight alternatives. UCSB presents that variety as a response to the range of research on campus: a computer science laboratory and a historian working through archival texts may need very different model characteristics.
The deployment has an earlier policy foundation. An August 27 memo from Executive Vice Chancellor and Provost David Marshall and Academic Senate Chair Rita Raley identified Claude, Qwen and Llama among the model families available through AI Commons. It also said faculty and staff retained Gemini access through the campus Google Workspace environment.
Chief Information Officer Josh Bright describes the approach as experimental, rather than a commitment to a fixed set of tools. In the October announcement, he argued that direct engagement is the best way to evaluate the technology, instead of watching from the sidelines or imposing rigid restrictions.
We are establishing a base camp in the rapidly changing world of AI rather than locking ourselves into a permanent destination.
Josh Bright, UC Santa Barbara chief information officer
Academic work stays under campus terms
The privacy protection is more specific than a general promise of secure access. According to the August memo, contracts bar vendors from training on or retaining academic and research material submitted through AI Commons, including unpublished work. That protection applies to the university-managed platform; the memo does not extend that statement to every AI service.
Marshall and Raley also tied model diversity to institutional control. Their memo says the campus should not depend exclusively on one vendor because models and companies change quickly. It argues that the university should set the terms governing academic work, rather than accept a provider’s defaults.
The memo’s governance principles separate software access from authority over academic decisions. Two requirements sit alongside the technical deployment:
- Academic oversight: The Senate and Academic Affairs should oversee teaching, scholarship, evaluation and intellectual integrity, even when IT and administrative offices handle implementation.
- Critical literacy: Equitable access should support informed, ethical engagement with AI, not substitute for learning how to assess and use it.
Access details also have a timing distinction. The August memo explicitly described AI Commons access for faculty and staff, with subsequent student access expected to come with academic-integrity guidance. October’s announcement describes campuswide rollouts but does not give a separate student-access schedule.
Employee training has an October 14 deadline
UCSB’s training effort includes AI@Work, a free certificate program for employees sponsored by Human Resources, Information Technology Services, and Professional and Continuing Education. It brings together around 30 participants over nine weeks, combining instruction and discussion with hands-on training and practical application.
The first cohort began in spring 2026. Employees can submit self-nominations for the second cohort through October 14, 2026. The program is an existing part of the university’s AI-literacy work, rather than a course newly created for the October deployment announcement.
The university’s stated literacy goals extend beyond employees learning which buttons to press. UCSB says students, faculty and staff should understand how AI systems function, recognize their biases and use them responsibly. Bright places that ethical framework alongside software functionality as a continuing campus priority.
Reader comments
Newest comments first. Replies stay oldest first.