Human-centered
AI supports — it does not replace — human judgment in teaching, assessment, and decisions about people.
AI policies & guidelines
The rules that govern AI use at Meridian, in plain language. Each policy below links to the authoritative document and carries a visibility label.
AI principles
PublicMeridian's commitments for adopting AI responsibly across the university.
Transparent
The community can see what AI tools exist, who owns them, and what data they may handle.
Private & secure
AI use follows the university's privacy, security, and data-classification requirements by default.
Equitable & accessible
Approved tools are evaluated for accessibility and offered broadly so benefits are shared across campus.
Acceptable use
Institution-wideWhat you may and may not do with university AI tools.
You may
- Use approved tools for coursework, research, and university work
- Enter Public and Internal information
- Use AI to draft, summarize, analyze, and learn
- Build on the approved API gateway with review
You may not
- Enter Confidential or regulated data into general tools
- Use AI to make final decisions about people without review
- Bypass the gateway to send data to unapproved services
- Misrepresent AI output as independently verified fact
Academic integrity & AI
PublicAI use in coursework is set at the course level. These are the common patterns instructors choose from.
| Course stance | What it means | Student action |
|---|---|---|
| AI not permitted | No AI assistance on graded work | Do not use AI tools for the assignment |
| AI permitted with disclosure | AI allowed if you say how you used it | Add a brief AI-use note to your submission |
| AI permitted freely | AI use is an expected part of the work | Use AI as a normal tool; verify outputs |
| AI required | The assignment is about working with AI | Follow the prompt's specific instructions |
The syllabus governs. When a course policy and a tool's capabilities disagree, the course policy wins. If a syllabus is silent, ask the instructor before using AI on graded work.
Data classification for AI use
Institution-wideThe single most important question before using any AI tool: what kind of data am I putting in?
| Classification | Examples | Where it's allowed |
|---|---|---|
| Public | Course catalog, published research, public web text | Any approved AI tool |
| Internal | Draft documents, internal email, non-sensitive operations | Meridian GPT, Copilot, approved suites |
| Confidential | Student records, PII, financial or health data | Only reviewed business systems or the Research Cluster |
| Regulated | HIPAA, FERPA-protected, export-controlled data | Secure Research AI Cluster, by agreement only |
Not sure how something is classified? Treat it as Confidential until you confirm, and contact the Information Security Office.
NIST AI RMF profiles
Restricted InternalMeridian adapts the NIST AI Risk Management Framework into context-specific profiles. Each profile tailors the framework's four functions — Govern, Map, Measure, Manage — to a particular setting, and guides how services in that setting are reviewed and approved.
The profiles split across the two governance lanes: most govern how AI is used in a context (Teaching, Research, Administration, Student Engagement), while Agentic & Cloud AI governs the shared platform itself — the AI API gateway behind Meridian GPT.
| Profile | What it tailors | Owner | Visibility |
|---|---|---|---|
| Teaching & Learning | AI in instruction, assessment, and academic support | AI Governance Committee + Provost | Restricted |
| Research & Scholarship | Sensitive-data research, reproducibility, secure environments | Research Computing + IRB | Restricted |
| Administration & Operations | Staff productivity, records, business-system AI | Office of the CIO | Restricted |
| Student Engagement | CRM, advising, and communications AI | Enrollment Management | Restricted |
| Procurement & Vendor AI | Due diligence, contracts, and vendor assurance | Procurement + Information Security | Restricted |
| Agentic & Cloud AI | Autonomous agents and the AI API gateway | Enterprise Architecture | Restricted |
Public summaries, internal detail. Full profiles are Restricted Internal because they document specific risk treatments. Meridian publishes a short public summary of each so the community understands the expectations that apply to its setting.