Artificial Intelligence (AI) Policy

Effective Date: January 27, 2025        

Purpose

The purpose of this policy is to provide guidance to all Toronto Public Library (TPL) staff for the secure, responsible, ethical and effective use of AI tools. The policy outlines the requirements that all staff must follow when using AI tools in the workplace.

Scope

This policy applies to all TPL staff using public AI tools, including AI functions embedded within existing software, and any private AI systems that TPL develops or purchases for TPL business purposes. The policy applies to the use of AI tools in all of TPL’s service pillars and administrative functions, including programming, collections, technology, data analysis, human resources, and procurement.

TPL will work to ensure that all external partners, vendors and presenters adhere to this policy and incorporate the responsible and ethical use of AI tools in their dealings with TPL.

The individual customer use of AI tools is out of scope of this policy. The TPL Rules of Conduct policy, Public Computing & Internet Use policy, and Wireless Terms and Conditions apply to all customer activities, including those undertaken using AI tools provided by or accessed through resources provided by TPL.

Underlying Principles

This policy is driven by principles as defined in the Library’s Values, Vision and Mission and in TPL's Equity Statement and Intellectual Freedom Statement.

Policy Statement

Toronto Public Library recognizes the transformative potential of AI tools to enhance library operations and services. They have the potential to enhance access to knowledge, provide valuable insights, improve efficiency and spur innovation. However, the use of AI tools also presents challenges and risks to TPL operations related to information security, privacy, transparency, equity and disruption to information environments.

TPL is committed to providing staff with the tools and information they need to successfully engage with artificial intelligence platforms and the content that they generate. This policy sets out principles for the secure, responsible, ethical and effective use of AI tools by staff, while mitigating the risks involved. This policy aligns with relevant legislation, frameworks and leading practices related to the ethical and transparent use of AI. It outlines TPL’s commitment to responsible AI use to ensure ethical considerations are upheld, AI risks are managed and compliance to emerging regulation is achieved.

  1. Security and reliability

    • 1.1 TPL is committed to responsible deployment of AI tools that strive to achieve appropriate levels of the trustworthy characteristics outlined by the National Institute of Standards and Technology AI Risk Management Framework (NIST AI RMF).
    • 1.2 An IT Security Assessment will be completed before an AI tool is approved and/or deployed for use by staff and appropriate risk mitigation measures identified and implemented.
    • 1.3 TPL staff will only use AI tools that have been assessed and approved before use to ensure they are safe, secure and robust, and reliable for the specific purpose intended.
    • 1.4 All TPL IT security practices will be applied to the use of AI tools, consistent with TPL’s Information Security Policy.

  2. Privacy

    • 2.1 Recognizing that uploading or inputting data into some AI tools may provide data to third parities for their own use and/or contribute to the training of the AI model, TPL is committed to minimizing the impact of this risk. Assessment and understanding of how the AI tool will store, transmit, and use inputs will be incorporated into the review of recommended AI tools.
    • 2.2 An AI Privacy Impact Assessment will be completed before an AI tool is approved and/or deployed for use by staff and appropriate risk mitigation measures identified and implemented.
    • 2.3 When using AI tools, all TPL confidentiality and data privacy practices will be upheld, in accordance with TPL’s Access to Information and Protection of Privacy Policy and the Municipal Freedom of Information and Protection of Privacy Act (MFIPPA).
    • 2.4 At a minimum, Personal Information (PI) and other restricted and confidential information will never be uploaded into an AI tool, unless the tool has been specifically approved by TPL for that purpose. Staff will not input data (including any confidential information), text, or pictures that would compromise the TPL.

  3. Accountability and transparency

    • 3.1 TPL is committed to ensuring that all processes and content related to AI are subject to proper oversight mechanisms and to providing transparency about the use of AI.
    • 3.2 When using AI tools, TPL staff should recognize the limitations and biases of AI technologies and critically evaluate the outputs. AI tools can enhance productivity but do not replace professional judgement or decision making.
    • 3.3 TPL staff are responsible for the accuracy and appropriateness of AI-generated content and will review outputs for accuracy as well as compliance with organizational standards and values. No AI-generated work product will be used or published without oversight from a TPL staff person. Decisions impacting peoples’ lives should not be fully automated and should always be made by humans.
    • 3.4 TPL is committed to transparency about where and how AI is used. Meaningful and timely information about use of AI tools by the Library will be provided publicly, that includes clear explanation in accessible plain language.
    • 3.5 TPL staff are responsible for clearly identifying all work products that have been substantially generated using AI.

  4. Algorithmic bias and equity

    • 4.1 TPL recognizes that algorithmic bias is a well-known risk associated with AI and is committed to using AI in ways that are non-discriminatory and promote inclusivity and equitable outcomes.
    • 4.2 A Human Rights AI Impact Assessment will be completed before an AI tool is deployed.
    • 4.3 TPL will strive to ensure that AI tools are tested for bias and that work products created through AI tools are vetted by staff subject area specialists for discriminatory bias.
    • 4.4 TPL staff are responsible for ensuring that the data chosen to be used in an AI tool for a specific purpose is of high quality, accurate and free of bias that can pose risks, cause harm, or reinforce various forms of discrimination.

  5. Sustainability

    • 5.1 TPL recognizes that generative AI systems can be a significant source of greenhouse gas emissions and water usage and will promote the environmentally responsible development and use of AI.
    • 5.2 TPL will strive to incorporate a sustainability assessment, based on available information, into the selection of AI tools, and aim to recommend AI tools that incorporate lower energy consumption approaches in the design of software, hardware and training models.
    • 5.3 Staff should be mindful of the extreme resource needs of AI tools and consider using an alternative approach if the use of AI for the same task would yield a comparable result and use of staff time.

  6. Continual Learning

    • 6.1 As the technological and regulatory environment of AI rapidly develops, TPL is committed to a culture of open-mindedness, flexibility, and dialogue. TPL will continue to engage with staff, partners, peers, stakeholders, and the public to invest in shared knowledge and a shared vision for responsible AI.
    • 6.2 TPL is committed to respecting copyright and will continue to monitor the legal status of AI tools in Canada as policies, regulations and case law evolves.
    • 6.3 TPL will provide ongoing training for staff on the use of AI in the workplace. This will include the cybersecurity and privacy implications of the use of AI in the workplace, as well as training on how to assist customers in understanding AI-generated content online and in media.
    • 6.4 AI has the potential to have significant impacts on library services, including related to data management, preservation, discovery, analytics and the need for AI literacy skills development. TPL will continue to monitor and assess the future impacts of AI on libraries, and adapt services as needed.

Specific Directives

Guidelines for application of this policy to specific service areas will be developed to provide direction to staff in the application and implementation of this policy.

Accountability

The Director, Policy, Planning and Performance Measurement has overall responsibility for this policy.

All Directors are accountable for ensuring implementation and compliance with the Artificial Intelligence Policy in their respective areas.

Related Legislation

Related Library Policies

Definition of Terms

Algorithmic discrimination/bias
Refers to instances when automated systems contribute to unjustified different treatment or impacts disfavoring people based on their actual or perceived race, color, ethnicity, sex (including based on pregnancy, childbirth, and related conditions; gender identity; intersex status; and sexual orientation), religion, age, national origin, limited English proficiency, disability, veteran status, genetic information, or any other classification protected by law.
Artificial Intelligence
A machine-based system that, for explicit or implicit objectives, infers from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.
Confidential Data
A class of data that should be contained within the organization to the extent possible as general disclosure could negatively influence operations and/or the TPL brand, but where specific disclosure of certain sub-sets may be permitted or required by law.
Generative AI
A type of AI that produces content such as text, audio, code, videos and images. This content is produced based on information the user inputs, called a “prompt,” which is typically a short instructional text.
Personal Information
Recorded information about an identifiable individual, including:
  1. information relating to the race, national or ethnic origin, colour, religion, age, sex, sexual orientation or marital or family status of the individual,
  2. information relating to the education or the medical, psychiatric, psychological, criminal or employment history of the individual or information relating to financial transactions in which the individual has been involved,
  3. any identifying number, symbol or other particular assigned to the individual,
  4. the address, telephone number, fingerprints or blood type of the individual,
  5. the personal opinions or views of the individual except if they relate to another individual,
  6. correspondence sent to an institution by the individual that is implicitly or explicitly of a private or confidential nature, and replies to that correspondence that would reveal the contents of the original correspondence,
  7. the views or opinions of another individual about the individual, and
  8. the individual’s name if it appears with other personal information relating to the individual or where the disclosure of the name would reveal other personal information about the individual.
Private AI
An AI system that is developed or acquired by an organization, and operates in a closed, restricted environment.
Public AI
AI system that is accessible to a broad audience through open platforms, APIs, or cloud-based services. Data input into public AI systems becomes part of the public dataset and is therefore accessible to other users.
Restricted Data
A class of data that must be contained within the organization to only those staff, volunteers and third parties working on behalf of TPL who need to know the information to perform their duties (including but not limited to personal information), as any disclosure not expressly permitted or required by law could not only negatively influence operations and/or the TPL brand but could also constitute a breach of law or contractual or regulatory requirements, and put TPL, external organizations and/or affected individuals at risk of legal, financial and/or reputational damage
Work product
Any deliverable or tangible outcome produced as a result of work activities within a project or task. This can include documents, emails, software, presentations, reports, designs, models, and other artifacts that demonstrate progress or completion of work, measure performance, ensure quality, or facilitate communication among stakeholders. At TPL, this could also be branch posters, social media posts, programming content, website content, etc.

Contact

Director, Policy, Planning and Performance Management

Print this page