Data
The campusGenAI platform infrastructure is approved for what UMass Amherst InfoSec has categorized as Level 3 data, including:
Non-directory information covered under FERPA, including but not limited to student grades, class schedule, and degree progress.
University personnel and financial records.
Any research data categorized as Level 3 by an Institutional Review Board, data steward, or data use agreement.
Please note that demo.campusgenai.org, as well as customized demos (before full adoption) are only approved for Level 2 data. Please also note that the Level 3 data categorization was determined and reviewed by UMass Amherst InfoSec. We recommend that other institutions consult their own InfoSec term or department for your security level.
See Data Categorization Levels
Privacy
For the most current and up-to-date privacy information, see UMass Amherst’s Privacy Statement. Please note that all academic institutions should consult with their own security teams regarding their policies. The UMass Amherst Privacy Statement is being actively developed by the UMass Amherst Information Security team, and may not be completely applicable to your own institution’s policies.
Temporary chat feature
The Temporary Chat feature is available as a no-history, no-storage option for users. These chats are excluded from search results, cannot be bookmarked, and are automatically deleted after 30 days.
To open a temporary chat, click the temporary chat message at the top right of the platform interface.
Frequently asked questions
No, information shared with the platform is restricted to your account and not accessible by others using the platform. As with other university services, a small number of UMass developers are able to access information shared on the platform, but only do so if required in compliance with the Acceptable Use Policy, Information Privacy Policy, and Confidentiality of Information Policy of UMass Amherst.
No, each new chat starts fresh, with no memory of past conversations. If you reopen and continue an existing conversation, it will remember context from your conversation. No information entered in one user’s chat can be accessed, remembered, or used in another user’s chat (or even your own separate chats).
No. We are committed to making the platform a trusted space for exploration and experimentation with AI.
In rare circumstances, the university may be required to review data in compliance with the Acceptable Use Policy, Information Privacy Policy, and Confidentiality of Information Policy.
Currently, you can share up to Level 2 Data.
User files and chat histories are securely stored on our platform until they are deleted by users. Stored chat data is used solely to provide users with access to their saved chat history. It is never used to train models or profile users, and users can manually delete their data at any time. The Temporary Chat feature is available as a no-history, no-storage option for users. These chats are excluded from search results, cannot be bookmarked, and are automatically deleted after 30 days.
Azure and AWS Bedrock are platforms that act as “cloud providers” or “model gardens.” They offer a managed and secure way to access models developed by OpenAI, Anthropic, DeepSeek, and others. These platforms host the models and handle key infrastructure tasks like API access and authentication.
The actual model companies, like OpenAI (ChatGPT), Anthropic (Claude), DeepSeek, etc. develop and maintain the models themselves, but we access those models through Azure and AWS Bedrock rather than directly from the model company.
For more information, see the model providers’ security and privacy documentation:
