Responsible Use of AI at WashU

Protect What Goes Into AI | Take Responsibility for What Comes Out of AI | Understand the Broader Risks of Using AI

Protecting WashU’s intellectual property and protected data is a shared responsibility.

AI tools are becoming part of everyday work, from drafting and summarizing content to brainstorming ideas and working with documents and data. As their use grows, it’s important to understand how to use them responsibly and where to find the tools and guidance available at WashU.

1. Protect What Goes Into AI Tools

Protect WashU Data and Sensitive Information

Before using any AI tool, consider what information you’re sharing and whether it’s appropriate for that tool. Follow all legal, regulatory, and University policies when collecting or storing sensitive data.

If unsure whether your tasks or documents contain sensitive data, contact:

• HIPAA Privacy Office for Protected Health Information (PHI)
• Office of the University Registrar for Protected Student Information (FERPA)
• Office of Information Security for Personally Identifiable Information (PII)
• Human Research Protection Office for use in human research studies

Using AI with PHI

Protected Health Information (PHI) is covered by HIPAA and has its own requirements for use with AI.

Not all AI tools are approved for use with WashU PHI. Before using any AI tool with PHI, confirm that the tool is approved for that purpose at ai.washu.edu/tools.

Approved AI tools aid judgment; they do not replace it.

• Tools must be on ai.washu.edu and approved for use with PHI. This site is updated often and has many additional details, including guidance on exception processes.
• AI tools are not a substitute for provider clinical decision making.
• Gemini Notebook, Claude Edu, and Copilot Chat are currently not approved for use with PHI.

APPROPRIATE USES OF AI TOOLS APPROVED FOR PHI

• Generate a denial letter for an insurance company.
• Care provider seeking general clinical info.
• Analyzing de-identified or confidential data with approved tools.

NOT APPROPRIATE USES OF AI TOOLS APPROVED FOR PHI

• Use of AI in lieu of clinical judgment.
• Copy/paste of unreviewed content into the medical record.
• Unapproved tool to conduct business or research with regulated, protected, de-identified or employee data.

Using AI with Copyrighted and Licensed Materials

Many materials used for teaching, research, and coursework, such as journal articles, e-books, textbooks, and datasets, are protected by copyright or by library license terms. Those terms can affect whether and how the materials can be shared with an AI tool.

For everyone

• Check the terms first. Library-licensed content (articles, e-books, databases) often has restrictions on use with AI tools. If unsure, ask a librarian.
• Use only what you need. Share the smallest portion necessary rather than full articles or chapters.
• Disclose AI use where appropriate. Follow the expectations of your instructor, supervisor, department, or school.

For students

• Ask before uploading instructor materials. Lecture slides, exams, problem sets, and course readings may be protected, so get your instructor’s permission first.
• Follow your course’s AI policy. Your syllabus or assignment may limit or prohibit AI use, regardless of copyright.

For faculty and staff

• Using AI with Copyright-Protected Data at WashU, a course available in Workday Learning, provides guidance on evaluating license terms and using protected materials responsibly.
• Be mindful when sharing materials with students. Consider whether content you provide to students, or ask them to use with AI tools, is permitted under its copyright or license terms.

Use the Appropriate AI Environment

Start with the tools WashU provides. WashU offers several enterprise AI tools, with options for a range of needs:

• General-purpose assistants such as Gemini and Copilot Chat support a wide variety of tasks.
• Grounded tools such as Gemini Notebook and Secure Ask WashU work from uploaded sources (intentionally limited).
• Built-in tools such as Box AI and Zoom AI Companion bring AI capabilities into applications you already use.

Before purchasing another AI service or turning to a personal account, visit ai.washu.edu/tools to see what is already available.

Keep WashU work in WashU enterprise AI environments. WashU work should only be conducted in WashU-approved AI workspaces, which provide enterprise-level protections that are not available with personal accounts or licenses.

Check the recommendation status of other AI tools. The AI Tools @ WashU page provides a comprehensive list of AI tools and their recommendation status, as reviewed by the Office of Information Security, including guidance on their use with protected data.

2. Take Responsibility for What Comes Out of AI Tools

Review and Verify AI-Generated Content

A polished response isn’t necessarily an accurate one. AI-generated content can contain errors, omit important context, or present incorrect information confidently. Apply your own judgment and verify important information before using or sharing it.

Before using or sharing AI-generated content:

• Check for errors: Verify facts, figures, and citations against reliable sources. AI can invent details that sound convincing.
• Look for what’s missing: Consider whether the output leaves out important context, perspectives, or nuance, or reflects bias.
• Review it yourself: Read the output closely and make the final decisions rather than relying on AI to do it for you.
• Disclose AI use when required: Follow the expectations of your instructor, supervisor, department, or school.

3. Understand the Broader Risks of Using AI

• Be alert for AI-enabled phishing. AI has made it easier for malicious actors to create sophisticated and convincing scams at a greater scale. Continue to follow security guidance and report suspicious messages using the Phish Alert Button in Outlook or by emailing phishing@wustl.edu.
• Adhere to current academic integrity policies. Students remain subject to the Academic Integrity Policies, the Student Conduct Code, and syllabus or assignment-specific guidance.
• Follow each tool’s usage policy. AI providers publish guidelines for responsible use of their tools and may take action when those guidelines are violated, including issuing warnings, suspending accounts, or ending access.
• Know when access ends. Your access to WashU AI tools ends if you leave the university, violate a Responsible Use Acknowledgement, or if and when a pilot program ends.