skip to main content
Caltech
Information Management Systems and Services
IMSS  /  Services  /  Generative AI

Generative AI at Caltech

Guidance for the Responsible and Secure Use of Generative AI

Generative Artificial Intelligence (GenAI) provides new opportunities to support research, education, and administrative activities at Caltech. The Institute supports responsible and thoughtful use of these technologies while protecting Caltech information, research, intellectual property, and the privacy of our community.

Caltech provides access to selected GenAI services that are managed or approved for institutional use and configured with security, privacy, and data protection in mind.

What You Need to Know

For Caltech work, use Caltech-managed, contract-approved, or SSO-protected GenAI services appropriate for your data and use case.

Do not enter confidential, restricted, regulated, or sensitive Caltech information into personal or consumer GenAI services.

  • Student records and FERPA-protected information
  • HIPAA-regulated information
  • Personally identifiable or sensitive HR information
  • Unpublished research, intellectual property, and proprietary datasets
  • Export-controlled information
  • Confidential financial or business information
  • Information subject to contractual, sponsor, data-use, or other research restrictions

Security Is a Shared Responsibility

Caltech and its technology providers implement safeguards for approved GenAI services. Users remain responsible for understanding the sensitivity of their information, selecting an appropriate service, following applicable requirements, protecting access, and reviewing AI-generated content before relying on or sharing it.

Important: Availability or approval of a GenAI service does not mean that every type of data, integration, or use case is approved for that service.

Guiding Principles

Caltech's guiding principles for responsible GenAI use are:

  1. Disclosure: When using GenAI, disclose or acknowledge its use when appropriate and follow requirements established by instructors, publishers, journals, sponsors, or other applicable authorities.
  2. Data and Information Protection: Protect Caltech information and follow applicable laws, regulations, contracts, and Institute requirements. Do not enter sensitive, confidential, or restricted information into open or consumer GenAI tools.
  3. Content Responsibility: GenAI systems can produce inaccurate, misleading, incomplete, or fabricated information. Review and assess GenAI output before relying on it, sharing it, or distributing it publicly.
  4. Caltech Honor Code: Use of GenAI remains subject to Caltech's Honor Code: "No member of the Caltech community shall take unfair advantage of any other member of the Caltech community."

Understanding GenAI Connections and MCP

Model Context Protocol (MCP)

Figure: Model Context Protocol (MCP) can connect AI applications with external tools and data sources. Security depends on how each connection, permission, and connected system is implemented and managed.

Model Context Protocol (MCP) provides a standardized way for AI applications to connect with external tools and data sources. Modern GenAI services can extend beyond a traditional chatbot by connecting to files, databases, APIs, applications, messaging platforms, analytics systems, and other tools.

These connections can expand the information and capabilities available to an AI application and therefore require additional security consideration.

Caltech approval of a GenAI service does not automatically constitute approval of every MCP server, agent, connector, plug-in, tool, or data source available through that service.

Only Caltech-approved integrations should be used to connect GenAI services to Caltech information, applications, or systems. Do not enable a connection involving sensitive or restricted information unless the integration and intended use have been appropriately reviewed and authorized by the IRB/Research Compliance Office and Caltech Information Security.

Before You Use GenAI

  1. Am I using a Caltech-enabled or otherwise approved GenAI service?
  2. Is the service appropriate for the information and use case involved?
  3. Does the GenAI service connect to other Caltech data, applications, tools, or systems?

If you are unsure, do not upload the information or enable the connection until the intended use has been reviewed.

Caltech AI Tools and Requirements

Learn more about the GenAI services available through Caltech and which services are appropriate for institutional use: Caltech AI Tools

For institutional use of LLM technologies, users must also follow the Caltech LLM Services Security & Governance Standard.

Need Help?

If you are unsure whether a GenAI service, integration, or use case is appropriate for Caltech information, ask before uploading data or enabling the connection.

General GenAI and technology questions: [email protected]
Research-specific requirements: [email protected]