Stanford AI Playground Now Approved for High Risk Data

University IT’s AI Playground is now approved for use with high risk and sensitive data including personally identifiable information (PII). 

Personal health information (PHI) such as patient data, and payment card industry (PCI) data such as credit card information, are under review but still not approved. 

Examples of high risk data include:

• HR and personnel documentation

• employee tax forms like W-2s or I-9s

• financial aid records

• salary or compensation data

• procurement and vendor contracts with sensitive pricing terms

• accommodation requests 

What this means for you

This approval covers all currently available AI models in the Playground, including OpenAI, Google, Anthropic, and Meta. The platform remains protected by Stanford single sign-on, and as before, your data stays within the Stanford environment.

Your school, department, or business unit may have additional guidance or policies related to high-risk data. Please check in with your local administration for any supplementary guidelines.

Getting started

New to the Playground, or haven’t used it recently? Visit aiplayground.stanford.edu and sign in with your SUNet ID. View the Quick Start Guide and dive into Slack for community discussion and peer tips through the #ai-playground channel. 

For technical support, submit a Help request through UIT.

Be responsible

Working with high risk data requires responsible use; view the Information Security Office’s Risk Classifications. Human review of AI-generated outputs, careful attention to what you include in prompts, and awareness of your obligations under university data policies and other regulations remain essential. 

Stanford’s Responsible AI and Responsible Agentic AI guidelines are good starting points if you have questions about appropriate use.

When in doubt, contact the Information Security Office.