Best Practices for Using Generative AI at Barry University
Sensitive Data Guidelines
-
Data Privacy & Protection
- Golden Rule: Avoid inputting data about others you wouldn't want shared about yourself.
- Strict Prohibition: Never input any University Data (esp. PHI, PII, FERPA) into any AI tool without explicit ISO authorization.
- Public Models: Using public/unvetted AI models with University Data is strictly prohibited.
- Opt-Out: Where possible, opt out of allowing AI tools to use your inputs for model training.
- User Consent: If using AI to interact with users, obtain their informed consent regarding data use and provide opt-out/deletion options.
-
Authorization & Approved Tools
- Mandatory ISO Vetting & Approval: MUST get prior ISO authorization before using any AI tool with any University Data. Only ISO-vetted/approved tools are permitted.
- Purpose Limitation: Only use approved AI tools for purposes authorized by ISO and University policy.
-
Academic & Professional Integrity
- Transparency & Citation: Cite AI use where required/permitted by academic or professional standards.
-
Assist, Don't Replace: Use AI to assist, not replace critical thinking or original work.
-
Verification: Always critically evaluate and verify AI-generated facts or critical information.
-
Security & Compliance
- Adhere to Policies: Strictly comply with ACUP, WISP, Password Policy, etc.
- No Personal Devices/Storage: Do not use personal devices or personal cloud storage for University Data. Use only approved University systems (e.g., Barry OneDrive).
- Report Concerns: Immediately report suspected misuse, breaches, or policy violations to the Information Security Office (ISO) at iso@barry.edu.
- Mandatory Training: Stay current with required Information Security Awareness Training.
-
Ethical Considerations & Transparency
- University Principles: Align AI use with Barry's principles: Human-Centric, Transparent, Fair, Secure, Accountable.
- Promote Discussion: Engage in conversations about ethical AI use, limitations, and benefits.
- Awareness of Bias: Critically assess AI outputs for potential bias stemming from training data.