Artificial Intelligence (AI) Security Concern

AI technologies simulate higher level thinking and can be used to perform tasks that might typically require input from a person. Several online AI tools have recently become publicly available, with ChatGPT being the most well-known.  These services include natural language processing and machine learning in their architecture, which are utilized to interact with people and generate content.  They are “trained” on the data they consume and pull from this growing body of knowledge for subsequent interactions.  This means that protected and confidential information entered by an employee can be stored by the application and shared with others outside of the University.

UConn’s Data Classification Policy is the institutional safeguard that applies in this situation.  If housed in the cloud, protected and confidential data is required to be stored only on information systems managed or contracted by the University.  Sharing of this information with other online systems (e.g., opensource cloud solutions, free software-as-a-service offerings, etc.) is not permitted.  You can view the policy at: policy.uconn.edu/2012/06/21/data-classification-information-technology