Responsible Use of Artificial Intelligence (AI) Policy

Middletown Public Library’s Responsible Use of Artificial Intelligence (AI) Policy

Purpose:

The Middletown Public Library (MPL) recognizes that Artificial Intelligence (AI) tools may support staff productivity, improve efficiency, enhance services, and assist with innovation, problem-solving, and public education.
 

This policy establishes guidelines for the responsible use of AI by library staff, while prioritizing patron privacy, data security, professional judgment, and public trust.
 

AI tools may assist staff work, but do not replace human oversight, critical thinking, or professional responsibility. 
 

AI use shall not diminish patrons’ ability to seek assistance directly from library staff.

Use of AI by staff shall be consistent with the Library’s commitments to privacy, intellectual freedom, equity of access, and professional ethics.

This policy shall be reviewed at least annually and updated as technology, legal requirements, and best practices evolve.

Definitions:
Artificial Intelligence (AI):

Software or systems that generate, analyze, summarize, recommend, or otherwise assist with tasks through automated or machine-learning processes.

Personally Identifiable Information (PII):

Information that identifies or could reasonably identify a specific individual, either directly or indirectly.

Confidential or Protected Information:

Non-public information that is legally protected, operationally sensitive, or entrusted to the Library, including but not limited to patron records, personnel records, security information, legal communications, and certain financial or contractual information.

Acceptable Uses:  

Staff may use AI tools for low-risk professional assistance and operational support, including but not limited to:

• Drafting, reviewing, summarizing, or improving:
  • Emails
  • Reports
  • Policies
  • Job descriptions
  • Meeting minutes
  • Presentations
  • Grant proposals
  • Public communications
• Generating ideas for:
  • Programming
  • Marketing content
  • Naming or branding concepts
  • Training materials
• Assisting with:
  • Analysis of anonymized or non-sensitive operational data
  • Budget review or trend analysis, provided sensitive personal data is not entered
  • Strategic planning and problem-solving
  • Supervisory or management guidance, provided identifying details are removed
• Supporting library service through:
  • Assisting patrons with AI-related questions
  • Demonstrating AI tools for educational purposes
  • Developing AI literacy programming
  • Generating accessibility tools such as alt-text
     

This list is illustrative and not exhaustive.

Privacy and Confidentiality Requirements:  

Protection of patron privacy and confidential information is paramount.

Staff shall not enter confidential, personally identifiable, legally protected, or security-sensitive information into AI tools.

Examples include, but are not limited to:

• Patron borrowing or account records
• Confidential patron inquiries or reference transactions
• Names or identifying patron information
• Personnel records or employee disciplinary matters containing identifying details
• Passwords or credentials
• Security procedures or system configurations
• Attorney-client communications
• Non-public donor information
• Legally protected records
• Any information protected under applicable library or state confidentiality laws

When seeking assistance from AI tools on operational, supervisory, or strategic matters, staff should remove or anonymize identifying details whenever feasible.

Staff should consider whether uploaded documents may contain embedded metadata or other hidden identifying information.

Public or free AI tools may retain prompts or submitted information unless documented otherwise, and staff should use caution accordingly. 

When AI tools offer privacy controls, staff should enable settings that limit data retention or use of prompts for model training, when feasible.

Human Oversight and Professional Judgment:  

Staff are responsible for reviewing and validating AI-generated content for:

• Accuracy
• Completeness
• Bias
• Fabricated or misleading information
• Copyright or plagiarism concerns
• Alignment with Library policy, professional ethics, and public trust

AI-generated content shall not be relied upon without appropriate human review.

AI may assist decision-making, but shall not replace professional judgment.

Prohibited Uses:

AI shall not be used to:

• Substitute for human judgment in personnel or disciplinary matters
• Substitute for legal advice or make legal determinations on behalf of the Library
• Substitute for staff professional judgment in responding to reference questions
• Generate or distribute content known to be misleading, discriminatory, or unlawful
• Use tools with unclear privacy practices or inadequate security protections
• Create deceptive synthetic media intended to mislead the public

This list is not exhaustive.

Approved and High-Risk Tools:

• Staff may use publicly available or personally subscribed AI tools, provided such use complies with this policy
• Requests for adoption of library-wide AI tools shall be subject to administrative review
• The Library may restrict or prohibit specific AI tools based on privacy, security, legal, or operational concerns

Transparency:
When AI substantially contributes to information provided to patrons or the public, staff should exercise professional judgment regarding appropriate transparency.

Staff shall not represent unverified AI-generated content as authoritative.

Training:

Staff should receive periodic training on:

• Responsible AI use
• Privacy and confidentiality risks
• Verification and fact-checking techniques
• Bias and ethical concerns
• Appropriate anonymization practices
   

AI Literacy and Public Education:

The Library supports public education related to AI and encourages staff to provide programming, assistance, and resources that help patrons:
 

• Understand AI tools
• Evaluate risks and limitations
• Use AI critically and responsibly
• Access AI-related learning opportunities equitably
   

The Library does not guarantee the accuracy or reliability of outputs generated by AI platforms used or discussed in library services.
 

Incident Reporting:

Any suspected privacy breach, accidental disclosure, or inappropriate AI use shall be reported promptly to Library administration.
 

Records Retention:

Staff shall manage AI-generated work product in accordance with applicable state records retention requirements.

Approved by the Middletown Board of Library Trustees on:  Tuesday, April 29, 2026