Meridian State UniversityOffice of the CIO · AI Services
Institution-wide
AI Service Card
Meridian GPT
| Document type | AI Service Card | Version | 2.1 |
| Owner | Office of the CIO — AI Services | Approver | AI Governance Committee |
| Primary audience | Students, faculty, staff | Visibility | Institution-wide |
| Last reviewed | 2026-05-12 | Next review | 2026-11-12 |
1. Service overview
Meridian GPT is the university's secure, multi-model AI workspace. It lets the campus community chat with approved AI models, upload documents for analysis, and use shared prompt templates. Prompts and uploaded files are not used to train vendor models.
2. Intended users & access
Eligible users
All active MSU account holders (students, faculty, staff).
Authentication
University single sign-on (SSO) with multi-factor authentication.
Cost to user
None — centrally funded.
3. Approved uses
- Drafting, editing, and summarizing non-sensitive documents
- Brainstorming, study support, and coding assistance
- Faculty preparation of course materials and assignment design
4. Prohibited uses
- Entering Confidential or regulated data (SSNs, full student records, health data)
- Sole reliance on outputs for grades, admissions, or hiring decisions without human review
- Generating content that violates the Acceptable Use or Academic Integrity policies
5. Downstream tools & use governance
Permitted
Personal productivity use and sandbox-style experimentation (trying models, prompts, and tools on non-sensitive data), plus departmental chat assistants and shared prompt templates within the Internal boundary.
Conditionally permitted
API-based integrations and custom tools — allowed only after AI Services review and gateway allow-listing.
Out of scope
Automated student-facing decisions, and any tool handling data above Internal — routed to the Secure Research AI Cluster instead.
6. Data handling & classification
Maximum data class
Internal. Do not enter data classified above Internal.
Training use
Contractually excluded — inputs are not used to train vendor models.
Retention
Conversation history retained per the system card; access logs kept 90 days.
Treat all inputs as Internal. When in doubt about a data class, consult the Data Classification for AI Use standard before proceeding.
7. Limitations & risks
Outputs may be inaccurate, outdated, or biased and must be verified before use. The service is subject to availability windows and rate limits. Model behavior may change as underlying models are updated.
8. Support & ownership
Service owner
Office of the CIO — AI Services
Support channel
ai-help@meridian.example
9. Related documentation
- System Card — MSU AI Gateway (Restricted Internal)
- Model Cards — underlying models (Mixed visibility)
- Acceptable Use guidance · Teaching & Learning RMF profile
Review & approval
Service owner — Office of the CIO
Approver — AI Governance Committee
Privacy & security reviewer
Date approved