Use this page to compare generative AI tools and services available to the Wharton community, including who can use each tool, approved data use, and how to get access.
Before you use any AI tool: Use of any AI tool must follow the Wharton AI Guidance, Penn Generative AI Guidance, and the Wharton Data Classification Standard. When in doubt, contact your Wharton Computing strategic partner.
Data-risk tiers:
- ● Low (public data) Public data only.
- ● Moderate (Non-PII / Non-FERPA) Internal data that does not contain PII or FERPA.
- ● Moderate (PII / FERPA) May include PII and/or FERPA data.
- ● High (restricted data) Restricted data such as HIPAA, SSN, or similar.
-
-
See the Wharton Data Classification Standard.
Chat & Conversational AI
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
Access: Get access |
Enterprise ChatGPT for Wharton with enhanced privacy, enterprise-level security, and University data agreements. Content entered in the Wharton ChatGPT Edu workspace is not used to train OpenAI models. Includes access to the latest ChatGPT models, workspace-shareable custom GPTs, and image creation. | ● Up to Moderate (PII / FERPA) | Moderate (PII and/or FERPA) data requires Wharton ISO risk review. API access and key handling must follow Wharton's AI API Key Requirements. API access is available for faculty and staff; students need a sponsoring faculty or staff member. ChatGPT Agent is not currently enabled. |
|
 Microsoft Copilot Chat (Basic) Access: Learn More |
Web-based chat agent for Microsoft’s generative AI offering, including web-grounded assistance for creating text and images. | ● Up to Moderate (PII / FERPA) | Personal Microsoft accounts are not protected. You must be signed in with your University account for Penn protections. |
Available at no additional charge to anyone with a PennO365 account. |
Access: Learn More |
Paid Microsoft M365 Copilot license that adds AI assistance to Microsoft Office applications, including Excel, OneNote, Outlook, PowerPoint, Teams, and Word. It also includes Work search for querying across those applications and PennO365 work content. Model availability may include OpenAI and selected Anthropic models depending on Microsoft and tenant configuration. | ● Up to Moderate (PII / FERPA) | Paid add-on license. Personal Microsoft accounts are not protected. Some Microsoft offerings are managed by individual schools. |
Available for faculty and staff for a nominal monthly charge; contact your Strategic Partner. |
 Access: Learn more |
Web-based interface for Google’s AI models, available through Wharton Google Workspace. Gemini can process and create text, images, and audio, and can help brainstorm ideas, summarize complex topics, draft outlines, parse data, and plan work. | ● Up to Moderate (Non-PII / Non-FERPA) | No PII or FERPA data. |
Full-time Wharton faculty, full-time Wharton staff, and Wharton PhD students with Wharton Google Workspace accounts. Access is by request only and is not automatically granted. Contact your Strategic Partner for access. |
Coding & Development
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
 Access: API key requirements |
Direct API access to OpenAI models for research, custom applications, integrations, and batch inference for large asynchronous workloads. API use must follow Wharton’s AI API Key Requirements, Penn/Wharton data policies, and any applicable DUA, contract, IRB, or other project-specific obligations. | ● Up to Moderate (PII / FERPA) | No High data. Moderate (PII and/or FERPA) data requires Wharton ISO risk review of the overall initiative’s data architecture and flow before production. Projects connected to Wharton-managed systems classified as Moderate or High also require risk review. Use must follow Wharton's AI API Key Requirements. API keys must be securely stored, never embedded in public repositories, and deactivated when no longer needed. Data generated or managed through the API must be stored only in University-approved cloud storage such as Penn Box, Dropbox, or OneDrive.
Batch-mode OpenAI API workflows are covered under this service.
Report security, privacy, or misuse concerns to security@wharton.upenn.edu. |
Faculty and staff with active PennKeys whose academic or professional responsibilities require API access. Students need a sponsoring faculty or staff member who funds the work and accepts responsibility for the risk. |
 Access: Request access |
Local code-focused agent access through ChatGPT Codex in the CLI, IDE extension, or app. Features can support code editing, debugging, Git workflows, and terminal-based task execution in a user's local development environment. | ● Up to Moderate (Non-PII / Non-FERPA) | Code only. No secrets, passwords, API keys, PII, FERPA data, High data, sensitive institutional data, or non-code Moderate (Non-PII / Non-FERPA) data. Codex Local is powerful: it can inspect and edit code, run terminal commands, and interact with repositories in a user's local development environment. Moderate (PII and/or FERPA) data or anything above Moderate (Non-PII / Non-FERPA) requires ISO engagement and risk review. Interconnections with other systems, DUAs/contracts, compliance obligations, or data pulled from Moderate (Non-PII / Non-FERPA) systems also require ISO engagement, including review of how data is protected in transit. Users must monitor agent actions and report information security or privacy concerns, unauthorized access, or suspicious behavior to security@wharton.upenn.edu. |
Disabled by default. To obtain access to ChatGPT Codex CLI, a user must be faculty, Wharton Computing full-time staff, or a Wharton PhD student working with faculty on approved code-only use cases. Users should already have a UPenn ChatGPT account and must provide written documentation attesting that they will adhere to the security requirements before access is enabled.
Broader distribution is still being evaluated and depends on Wharton security requirements, such as training, logging, and monitoring. |
Access: Request access |
Cloud-hosted code-focused agent access through ChatGPT Codex. Codex Cloud works in OpenAI-managed sandbox environments that can inspect repositories, propose edits, run supported commands, and prepare code changes without direct access to a user's local machine. | ● Up to Moderate (Non-PII / Non-FERPA) | Code only. No secrets, passwords, API keys, PII, FERPA data, High data, sensitive institutional data, or non-code Moderate (Non-PII / Non-FERPA) data. Because work runs in OpenAI-managed sandbox environments, Codex Cloud is more isolated than Codex Local for many development workflows. Tradeoffs include limited tools, limited access to local or internal systems, and slower execution for some tasks. Users must monitor agent actions and report information security or privacy concerns, unauthorized access, or suspicious behavior to security@wharton.upenn.edu. |
Available to Wharton users with approved ChatGPT Codex access for code-focused use cases.
Contact your Strategic Partner or support@wharton.upenn.edu to confirm eligibility and access. |
 Access: GitHub at Penn |
AI coding assistance from GitHub for supported IDEs, terminals, GitHub workflows, and developer tools. Copilot can suggest code, explain code, help with debugging, support code review workflows, and use repository or editor context depending on configuration. | ● Up to Moderate (Non-PII / Non-FERPA) | Code only. No secrets, passwords, API keys, PII, FERPA data, High data, sensitive institutional data, or non-code Moderate (Non-PII / Non-FERPA) data. Use the ISC-managed GitHub Copilot pathway for Penn/Wharton work. Do not use personal GitHub Copilot plans for institutional code or data unless separately approved. Moderate (PII and/or FERPA) data or anything above Moderate (Non-PII / Non-FERPA) requires ISO engagement and risk review. Users must monitor tool output and report information security or privacy concerns, unauthorized access, or suspicious behavior to security@wharton.upenn.edu. |
Available through ISC-managed GitHub Copilot licensing.
Wharton users should contact their Strategic Partner to request access. |
 Access: Request an AWS account |
Programmatic access to approved large language models hosted in AWS Bedrock for research workloads, custom applications, integrations, and batch inference. Bedrock complements the OpenAI API Platform by offering approved model providers available through AWS. | ● Up to Moderate (PII / FERPA) | Bedrock is not approved for High data by the University. Bedrock Serverless model terms must be approved by University Procurement before models are made available; previously available models may be revoked if terms change. Approved model providers/resellers include Anthropic models (all), Cohere, Mistral Models, and Twelve Labs. |
Wharton AWS account owners and users with approved AWS access.
AWS accounts identified as handling High data are not eligible for AWS Bedrock Serverless offerings. |
Meetings & Collaboration
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
Access: How to enable |
Generative AI features in PennZoom for meeting questions, meeting summaries, smart recordings, and whiteboard content generation.
Zoom AI Companion does not use meeting data, attendees, transcripts, or shared items to train AI models. |
● Up to Moderate (Non-PII / Non-FERPA) | Not approved for meetings involving PII, FERPA data, HIPAA/PHI, Social Security Numbers, credit card data, non-student minors, or other High data. All AI Companion features are off by default and must be started by the meeting host or co-host. Meeting summaries start only after the host clicks Start Summary and are shared by email, which can be forwarded. Summaries, recordings, transcripts, or generated materials that contain Moderate (Non-PII / Non-FERPA) data must be stored only in approved Moderate environments such as Penn Box, Dropbox, or OneDrive. |
PennZoom hosts and co-hosts can start AI Companion features when available in PennZoom.
Access may change pending University negotiations with Zoom. Participants cannot start or stop these features and are notified when they are enabled. |
Course & Classroom
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
| Lecture Recall (Faculty) Access: Pilot info |
A custom GPT inside ChatGPT Edu that helps turn recorded class sessions, slide decks, and practice materials into a course-specific AI study assistant for students. Currently running as a pilot in selected courses. | ● Up to Moderate (PII / FERPA) | No High data. Inherits ChatGPT Edu policy. Moderate (PII and/or FERPA) data requires Wharton ISO risk review. |
Wharton faculty teaching an upcoming Wharton MBA or WEMBA course selected for the pilot. The Lecture Recall setup process requires advance notice before course launch, access to class recordings, and TA support is recommended. |
| Lecture Recall (Students) Access: How to use |
A course-specific generative AI study assistant inside ChatGPT Edu for revisiting, searching, and quizzing yourself on recorded class sessions, slide decks, and practice materials. Students should cross-check important answers against the linked course materials or with the teaching team. | ● Up to Moderate (PII / FERPA) | No High data. Inherits ChatGPT Edu policy. Moderate (PII and/or FERPA) data requires Wharton ISO risk review. |
Students enrolled in eligible pilot courses. |
Research Computing & ML Platforms
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
| PARCC Betty HPC + AI Cluster Access: PARCC Betty |
PARCC is Penn's university-wide research computing center. Its Betty system provides centrally managed high-performance computing and AI infrastructure, including CPU and GPU resources for demanding research workloads. | ● Under Review | Data classification guidance is under review. Please consult with Wharton Research Computing and ISO through the risk review process before using PARCC for institutional or sensitive data. PARCC is owned by Penn's Office of the Vice Provost for Research. Wharton access is managed by the Research Computing team, the same group that supports access to HPC3. |
Penn research groups with approved PARCC access. Wharton researchers should email research-computing@wharton.upenn.edu to get started. |
| gRPt (Qualtrics + OpenAI) Access: Request via Research Programming |
Wharton Research Programming service for adding an OpenAI-powered chat experience inside Qualtrics surveys. It provides a generic chat front end and secure backend so study participants can interact with a model during a survey, with conversation history captured as survey data. | ● Up to Moderate (Non-PII / Non-FERPA) | Not approved for Moderate (PII and/or FERPA) or High data. Studies involving PII, FERPA, Moderate (PII and/or FERPA) data, or higher-risk data flows require Wharton ISO risk review and Research Programming review before production. Email research-programming@wharton.upenn.edu to get started. |
Wharton faculty, Wharton PhD students, and sponsored research teams running Qualtrics-based studies. Provisioned by Research Programming. |
Writing & Productivity
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
Access: Request Grammarly |
AI-assisted writing assistant for grammar, spelling, punctuation, citation style formatting, tone, clarity, and style. Available as a browser extension, web service, desktop app, and mobile keyboard. | ● Up to Moderate (Non-PII / Non-FERPA) | Not approved for Moderate (PII and/or FERPA) data. High data, including PII, payment card data, and HIPAA, is generally not permitted. HIPAA usage is pending additional Penn Medicine Privacy and compliance review. Wharton is maintaining license costs for eligible existing users through June 2027; new or ineligible users may need to fund accounts from their own budgets. Contact Client Support Services at computing@wharton.upenn.edu or your Strategic Partner to request access. |
Wharton full-time faculty, staff, and PhDs are eligible for Wharton-subsidized accounts. Other Penn affiliates may be eligible through ISC but must provide their own budget code. |
Pilots & Under Review
Tools requested by community members or undergoing evaluation. Listed here so you know they're on the radar.
| Tool | Description | Approved Data Tier | Data Usage Notes | Available To |
|---|---|---|---|---|
Access: Contact Apps Support |
Wharton Enterprise Airtable is a cloud-based, low-code/no-code platform for building custom relational database applications and workflows. Airtable AI can help write formulas, analyze feedback, generate content, and summarize data inside approved Airtable workspaces. | ● Up to Moderate (Non-PII / Non-FERPA) | Not approved for Moderate (PII and/or FERPA) or High data, except for separately approved limited exceptions such as faculty photos. PII is not permissible in general at this time. Data stored in Airtable and access granted to Airtable workspaces must follow ISO disposition requirements and Privacy Office guidance. PennKey authentication is supported. Editors/creators require paid seats; viewers are read-only. |
Wharton Computing and Wharton staff use cases. Requester must be a Wharton Computing staff member and provide a developer or analyst resource responsible for building and supporting the Airtable workspace. |
Policies, Guidance & Training
| Resource | Description |
|---|---|
| Wharton AI Guidance | Wharton-specific guidance for using generative AI responsibly, including privacy, security, teaching, research, and approved-tool considerations for the Wharton audience. |
| Penn Generative AI Guidance | Penn's central guidance on responsible use of generative AI: transparency, accuracy, privacy, IP, HIPAA, and bias considerations. The starting point for any AI-related question at Penn. |
| Generative AI at Wharton | Wharton Knowledge Base landing page for generative AI articles, including approved tools, ChatGPT Edu, Copilot, Grammarly, prompting, and Lecture Recall resources. |
| Wharton Data Classification Standard | Wharton's canonical four-tier framework (🟢 Low / 🟡 Moderate (Non-PII / Non-FERPA) / 🟠 Moderate (PII and/or FERPA) / 🔴 High) for what AI tools can be used with which data. Required reading before submitting institutional data to any AI tool. |
| Wharton Policies, Standards & Guidelines | Wharton Information Security policies, standards, and guidelines, including AI API Key Requirements, risk review, data classification, security awareness, and related Penn policy links. |
| Wharton AI Tools Hub | Wharton Computing's directory of approved AI tools, with current access details, security guidance, and best practices. The canonical source if anything on this page looks out of date. |
| Penn AI Training | Two tracks of self-paced AI training from ISC: General AI Literacy for every Penn affiliate, and a Technical / Developer track for builders. Funded through the Draw Down the Lightning grant. |
| CETLI - Generative AI & Teaching | Penn CETLI's hub for faculty: course policies, AI-aware assessment design, syllabus language, and academic integrity guidance. |
Questions?
Questions about a specific tool, or want to request a new one? Email support@wharton.upenn.edu or contact your Wharton Computing strategic partner.