AI Class Capture

Guideline for Academic/ Class Meetings:

Except for authorized lecture-capture hardware and software, these guidelines apply specifically to capturing academic/class meeting content by any automated or technical means including recording the meeting, using AI assistants (e.g., Zoom AI Companion, Otter.ai, Teams), or enabling transcriptions.

General Guidelines for Capturing Classes and Class Meetings 

  1. Nature and Purpose of Meeting 
    1. Instructors are advised to consider the nature of their classes and meetings before capturing them, including through services such as Otter.ai, Teams, or Zoom AI Companion.  
    2. Capturing of classes is allowed: 
      1. When the instructor is using enterprise tools such as Zoom and the distribution of recordings is limited to people who are enrolled in the class that is being recorded. This process meets those requirements when the instructor uses the integrations between Canvas, Zoom, and Kaltura to schedule, launch, and record the meeting.
      2. By students and instructors who have an authorized disability accommodation, and 
      3. By instructors using non-enterprise tools where students and attendees have been notified prior to the class. See notification and consent section below. 
    3. Capturing of meetings is prohibited for:  
      1. Classes and meetings where -risk (e.g., health information, identifiable human subjects research info, financial or loan information, undocumented student data) are discussed, or where medium-risk (e.g., attorney-client privileged dialogue, animal research protocols, student conduct deliberations) are discussed, or 
      2. Academic research purposes without risk assessments from the ISPO and approvals from the IRB, and the appropriate data steward, where necessary. 
    4. Unless called for by unique or rare circumstances, it is not advisable to capture meetings (e.g., office hours) where student concerns, behaviors, integrity issues, and other sensitive topics are discussed. These create unnecessary records and obligations related to them, take up storage space, expose individuals to significant privacy and security concerns, chill open communications, and may be subject to public records requests. 
    5. Deans, Directors, and Chairs may develop internal policies and procedures restricting the use of AI features, and likewise may limit use to only low risk data.
    6. Individuals are advised to use extreme caution when capturing meetings where other medium-risk personal or confidential information (e.g., research planning, assessments) is discussed.  
  2. Authorized Users 
    1. All users must adhere to the vendor terms of use, UNM policies, security and privacy guidelines and standards, and all laws and regulations when using AI Tools.  
    2. Students with an authorized disability accommodation may capture meetings in accordance with their accommodation. However, they are not authorized to distribute the output or use the output for any other purpose than their own personal accommodation needs. 
    3. Other than individuals with authorized disability accommodations, only the instructor/host is authorized to capture classes, unless the instructor chooses to allow students to capture classes (see B. 4 below). The host must review transcripts and summaries for accuracy in a timely manner before disseminating them to others.  
    4. Instructors may choose to allow students to “send” an AI assistant to “attend” a meeting or class on their behalf when the student is not present. However, no one may require students to “send” an AI assistant to “attend” on the absent individual’s behalf. Hosts may remove unauthorized AI assistants that are in classes without their users. 

 II. Privacy and Data Security Considerations

  1. Captured meetings are university records and may be subject to disclosure requests and may be subject to privacy and security requirements by policy and by law.  
  2. Captured meetings must be maintained at the appropriate security level for the sensitivity of the data included in them. 
  3. Output of captured meetings (e.g., videos, summaries) may not be saved on personal devices (i.e., non-UNM-issued) and must only be stored on university-approved services (e.g., in Kaltura (accessed via Canvas), not in one's personal accounts). 
  4. The retention period for captured meetings is one year from the date of capture unless it is necessary to maintain the information longer. The host is responsible for deleting the captured meeting at the end of the retention period. If, however, these captured meetings contain identifiable student data, then they must be retained for five years.
  5. Where a meeting will be captured, individuals must be provided the options to mute audio and turn off video. 
  6. High and medium-risk data (as specified above) may not be discussed, and students should be advised not to discuss sensitive information about themselves, while AI features are enabled or, when the meeting is being captured. 

III. Notification and Consent Requirements 

  1. Except in cases where advance notice is not possible due to an authorized disability accommodation, instructors must provide advance notice, where possible, to all participants that the class or meeting will be captured. Best practice for notice is to provide at least 2 of the following, along with an opportunity to voice concerns: 
    1. Inclusion of the information in the course syllabus (that meetings will be captured); 
    2. Reminder provided before the meeting or class with an opportunity to object; and 
    3. Reminder at the beginning of the meeting or class that it will be captured 
  2. Where there is an unauthorized or unknown AI Assistant present in the meeting or class, the host must remove it. 
  3. A student’s refusal to participate in a meeting with AI features enabled should not, by itself, be the subject of an adverse action (for example, courses enabling AI cannot require student participation as part of a lesson or course grade). Except where there is an authorized disability accommodation or the course involves AI assistant features as subject material, the host must disable AI features if a participant requests that features be turned off while presenting or asking questions. 

IV. Specific Considerations Regarding AI Tools: 

  1. Individuals may only use AI assistants that have
    1. Had risk assessments completed by UNM ISPO
    2. Had approvals from appropriate data stewards, where applicable
    3. Have a written agreement in place (i.e., contract, Master Services Agreement) that enforces UNM’s terms and conditions, as well as UNM’s privacy and security obligations.
  2. Some AI assistants collect and retain all information presented in meetings, including biometrics such as voices, which can be used to train the AI, or which can be shared without authorization outside the organization. These should not be used except where an authorized disability accommodation exists that non-AI tools cannot address