CFLA was pleased to bring together interested library association members from across Canada to discuss our submission to the federal government consultation on AI transparency, which we filed September 21, 2026. Thank you to all who contributed!
In October, we participated in consultation activities through Canadian Heritage and Amii related to AI, and we promoted the contributions libraries can make to the national AI strategy. We were pleased to participate alongside our library association partners CARL, OLA and FMD, and our colleagues from LAC and BAnQ, among others in the cultural sector.
Highlights from our submission to the AI transparency consultation are below and the full document is available on the CFLA website under Resources – see Responses to Consultations.
On detecting and identifying AI-generated content: The public needs to be empowered so that they are able to identify the signals of AI-generated content. Libraries need dedicated funding to develop programs to increase information and digital literacy among the public and students, to identify misinformation, disinformation and AI-generated content when it matters.
On identifying AI-generated content through metadata: AI-generated content should include both visible and invisible metadata. Visible metadata is necessary to reach the public. However, visible metadata alone is insufficient; invisible metadata is needed for system processing and in a business-to-business context. Libraries have used metadata for decades when purchasing, cataloguing and lending library materials, in standardized formats and we continue to collaborate internationally in this area.
Taking a tiered approach to identification: Identifying whether something is AI-generated is more important when users are making a decision that affects their lives or affects others, and less important when they are seeking entertainment. Information generated by AI that is likely to be used in the context of decision-making should be identified and accompanied with appropriate cautions about the accuracy and reliability of the information provided. It would be most effective if developers provided built in tools for users and consumers to identify and check if content is AI generated, such as with Google’s SynthID.
Explore using the existing ISBN and ISSN systems to require disclosure: we noted the role of LAC, and the problem of low-quality AI-generated books and serials that are flooding the marketplace. CFLA suggested that Library and Archives Canada receive resourcing to explore the potential to use the ISBN and ISSN systems in Canada as a mechanism to require disclosure of AI-generated content, and the need for the publishing industry’s participation in solutions. Note that after our submission, LAC announced that by applying for an ISBN, an applicant is declaring that the publication was written by a human author.
Legislative provisions must be protected from contractual override. Any regulation that arises over time from the government related to AI-generated content will have no impact if private contracts and terms of use can supersede the terms of the legislation. As Canada develops legislation and regulations associated with AI-generated content, it will be important to consider legislative terms that prevent contracts from over-riding the protections the government has established.
What’s next for responses to federal consultation? Watch for an invitation in early November to participate in focus groups for CFLA’s submission for Canada’s Voluntary National Review of the UN Sustainable Development Goals. You can read Canada’s 2023 report here, which spotlighted libraries under SDG 11 on page 72.