The Division of Information Technology and Analytics reminds the campus community that entering personal, confidential, or proprietary information into a Generative AI model without data protection is a risk.
Some examples of sensitive information are:
- Personal Identifiable Information (PII): Full names, addresses, social security numbers
- Student academic information: Grades, evaluations, or any other records that could impact student privacy and academic integrity
- Financial details: Bank accounts, credit card numbers, transaction data
- Health records: Personal medical information, health records
- Login credentials: Usernames, passwords, PIN, security questions
- Confidential academic work: Unpublished research, patents
If you use Generative AI models for your academic or professional purposes, please follow these data protection best practices:
- If you are Faculty or Staff, use Microsoft Copilot with your IC credentials. This enterprise tool provides data protection. For more information, please visit this article: https://help.ithaca.edu/TDClient/34/Portal/KB/ArticleDet?ID=1492
- Use anonymized, synthetic, or dummy data instead of real data when possible. Limit Data Sharing by only providing the minimum amount of data necessary for the tool to perform its function.
- Before using, review a tool’s privacy policy and terms of service to understand how your data will be used, stored, and protected. Give preference to tools that delete your data after processing.
- Understand and utilize data usage and retention settings that may be available.
- For example, ChatGPT has added the ability to turn off chat history and disable it’s use for training the model (it is still retained for 30 days). More information can be found here: New ways to manage your data in ChatGPT (openai.com)
Please submit a ticket with any questions here: https://help.ithaca.edu/TDClient/34/Portal/Requests/ServiceDet?ID=230
Thank you for your help in protecting the data of our community!