Understand the Types of AI Tools

AI tools vary widely in purpose and risk. Common categories include:

• Text based generative AI (e.g., tools for assisting with writing, brainstorming, summarizing, or coding)
• Image, audio, and video generation tools (e.g., tools that generate or modify images, voice, music, audio, or video)
• AI features embedded in enterprise software (e.g., AI assistants built into search engines, email, document editing, video conferencing, or learning platforms)
• Specialized or discipline specific AI tools (e.g., tools for research analysis, programming support, or instructional design)

Different categories of tools pose different risks depending on where your data goes, how outputs are used, and whether Tufts has a contract or data protections in place.

Seek Approval When Using Tools Not Pre-Approved

Faculty and staff must not use unapproved AI tools for Tufts business, instruction, research, or administration without review and approval, especially when institutional data is involved.

To request review or approval of a new AI tool:

• Follow the Technology Review and Procurement process
• Ensure the tool complies with Tufts security, privacy, accessibility, and contractual requirements

Be Accountable

You are responsible for all work, decisions, and outputs that involve AI, regardless of the tool used.

AI does not replace:

• Professional or academic judgment
• Scholarly responsibility
• Ethical obligations

If you submit, publish, share, or rely on AI assisted content, you are accountable for its accuracy, integrity, and appropriateness.

If an AI tool is accessing meetings or content on your behalf without your permission, contact TTS immediately for assistance removing or blocking it.

Practice Transparency

Transparency is expected when AI meaningfully contributes to:

• Academic work or assessment
• Research design, analysis, or writing
• Administrative or professional decisions
• Public‑facing communications

Follow disclosure expectations set by instructors, supervisors, journals, funders, or professional standards. When in doubt, disclose.

Be Aware of AI Tool Terms, Licenses, and Data Use

Before using any generative AI tool for Tufts related work, users should understand the vendor’s terms and conditions (T&Cs). These terms govern how data may be used, retained, disclosed, or incorporated into the vendor’s models—and may conflict with Tufts policies or legal obligations if not reviewed. AI tools differ significantly in how they handle user inputs and outputs.

When choosing or using an AI tool, users should be aware of whether the vendor’s terms allow:

• Training or fine-tuning on user inputs: Some AI tools reserve the right to use submitted content to train or improve their models. This may be incompatible with Tufts requirements when institutional, confidential, or sensitive data is involved.
• Retention and reuse of prompts or outputs: Vendor terms may permit storage, reuse, or human review of prompts and outputs—regardless of whether the tool is marketed as “free” or “private.”
• Broad licenses over outputs: Some providers assert licenses that allow them to reuse or distribute AI generated outputs, which may create conflicts with
  • Academic publishing requirements
  • Intellectual property expectations
  • Sponsored research obligations
• Restrictions on use cases: License terms may prohibit or limit use for
  • Commercial activities
  • Clinical or health‑related contexts
  • Regulated research
  • Automated decision‑making affecting individuals
• Lack of warranties or liability protections: Most AI tools disclaim responsibility for accuracy, bias, or infringement, placing full responsibility on the user and/or the institution.

• Students should understand that using AI tools does not transfer ownership, authorship, or responsibility and may affect originality, citation, or integrity requirements.
• Faculty should consider how AI tool licenses interact with publication agreements, grant terms, and scholarly norms.
• Staff should ensure that AI tool terms are compatible with Tufts’ data protection, procurement, and risk management requirements.

AI use is governed not only by Tufts policy, but also by vendor contracts that users are presumed to accept when using the tool. Because AI vendor terms can create legal, privacy, and compliance risks, Tufts requires that:

• AI tools used with institutional data be reviewed and approved through the TTS technology review and procurement process
• Users do not accept clickthrough agreements for AI tools on behalf of Tufts without authorization
• Personal subscriptions to AI platforms are not used for Tufts work unless reviewed and approved

Jump to "On This Page" Table of Content

Follow Tufts Data Handling Requirements

AI use must comply with Tufts’ Data Classification Policy. Data sensitivity determines whether an AI tool may be used.

Unless an AI tool has been explicitly approved for the data type, you must not enter:

• Restricted or regulated data, including:
  • Student education records (FERPA)
  • Health or clinical information (HIPAA)
  • Research participant or human‑subjects’ data
  • Personnel, HR, or payroll information
  • Legal, disciplinary, or security‑sensitive information
• Confidential or proprietary institutional data
• Contract‑restricted or licensed content

Assume that information entered into unapproved or public AI tools is no longer private and may be retained or reused outside Tufts’ control.

Additional Tufts AI Guidance and Governance

• Tufts Generative AI Usage Guidelines (Office of the Vice Provost for Education)
• AI Council and AI+ Initiatives
• AI Task Force and Institutional AI Strategy
• Educational Technology Services (ETS) – AI and Teaching Resources
• Center for the Enhancement of Learning and Teaching (CELT)
• Tufts Data Classification Policy
• Information Security Policies
• Academic Integrity Policies
• Research and IRB Guidance
• Accessibility Standards