The WashU ChatGPT Beta is compliant for use with sensitive data and WashU intellectual property, including information protected under HIPAA and FERPA.

The tool currently uses Open AI’s ChatGPT-4o model but, due to its isolated nature, does not pull in new information from the web.

Since its launch in December 2023, thousands from the WashU community have used the secure WashU ChatGPT Beta to enable and enhance research, education and patient care.

The tool has been made available in Beta form to quickly provide a secure GPT environment to the WashU community for research, operations, and education. It is not mobile-friendly, and users may experience limited capacity or constraints.

Users are invited to share chatbot feedback so that we may continue improving user experience and optimizing the tool.  Advanced users may also request API Beta access.

WashU ChatGPT Beta is approved for use with sensitive information, including data protected under HIPAA and FERPA.


WashU ChatGPT FAQs

Who can use WashU ChatGPT Beta?

WashU faculty, staff and students can access the sandbox using their WUSTL Key.

Why is this labeled a Beta? What can I expect during the Beta phase?

It is important to provide faculty, staff and students with swift access to this powerful emergent technology in a secure sandbox environment that protects sensitive data from leakage and where you can use your own data and work output without it becoming part of OpenAI’s model training. We have released our WashU ChatGPT instance while still in Beta in order to make a secure tool available to the university community as soon as possible. As a Beta, it may break or behave sub-optimally – a few hiccups are expected and we ask for your feedback and your patience as we continue to optimize the tool.

What version is this? How similar/different is this tool to the publically available version of ChatGPT? Is it trained on WashU-specific datasets?

WashU’s instance of ChatGPT uses OpenAI’s ChatGPT Version 4o. Currently, it is not custom-trained on any WashU-specific information or datasets. It does a fairly poor job at answering WashU-specific questions like library hours or HR policies and it is not suggested for these types of questions at this time.

Does information I enter become part of the model’s training, as happens with public versions of generative AI tools?

No, the secure sandbox will never display information you enter to another user, nor will it train itself with that information.

Can I submit information to become part of the model? Can I train my own custom model?

No, at this time we are not custom-training the WashU tool nor offering access to individual instances. Please provide feedback using the form at it.wustl.edu/gptfeedback or contact aiquestions@wustl.edu about the functionality you seek.

Does the tool pull in information from the internet in real time or only from a static model?

This tool is not able to search the live internet and does rely on a static model. If you need to rely on more updated content, and you are not working with sensitive information, please consider using Microsoft Copilot (formerly Bing Chat Enterprise), available to WashU students, faculty and staff, as that tool does have access to the live internet.

Is this tool approved for use with sensitive data like information protected under HIPAA or FERPA?

Yes, it has been reviewed by the university and is deemed suitable for use with sensitive data, including information protected under HIPAA and FERPA, and for any university purpose, including research, operations, teaching and learning. It is important to provide faculty, staff and students with swift access to this powerful emergent technology in a secure sandbox environment that protects sensitive data from leakage and where you can use your own data and work output without it becoming part of OpenAI’s model training.

How often will WashU’s instance be upgraded to the latest version?

The WashU GPT will consider each new OpenAI model release before implementing it locally. New model releases will not instantly be available in the WashU GPT.

Can I access the sandbox from my mobile device?

Technically, yes, you can, but the tool currently does not utilize a responsive design to make it usable on small mobile devices. You will need to be in a desktop environment to actually utilize the tool.

Does WashU track and retain data on individuals’ usage of the tool? What happens to deleted chats?

Individuals’ usage is logged in conjunction with their WUSTL Key login. All chat history is retained in the system even when deleted from display by the user. The tool exists to serve the missions of the university and its misuse or abuse will not be tolerated.

What kind of guardrails are in place around explicit langauge and dangerous content?

Content filtering is in use. You should refrain from participating in harmful or offensive activities. See WashU compliance policies wustl.edu/about/compliance-policies for more. All interactions are logged and linked to the user’s WUSTL Key account.

Does the Washington University ChatGPT Beta accept image input or produce image output?

No image-related capabilities exist with this tool – input and output are wholly alphanumeric.

Will WashU be adding other generative AI tools to its secure environment?

Yes, API endpoint access is now available and more announcements are planned. Please direct further inquiries to aiquestions@wustl.edu.

Will WashU be adding other generative AI tools to its secure environment?

Potentially. We will keep the campus community updated about further plans.

What is generative AI? How can I learn more about these emerging tools and technologies?

Please browse our resources to learn about AI tools and responsible use.