The Centre for Animal-Aligned AI Submits Recommendations on Animal Inclusion in Canada’s AI Transparency Framework
The Centre for Animal-Aligned AI (CAAI) has submitted a written response to Innovation, Science and Economic Development Canada (ISED) as part of its consultation on Advancing AI Transparency in Canada, calling for non-human animals to be explicitly considered within emerging AI transparency and incident-reporting frameworks.
The federal consultation is examining how Canada should approach transparency as AI systems become increasingly integrated into society. ISED has identified five areas for potential action, including disclosure of AI-generated content, transparency when people interact with AI systems, information about AI capabilities and limitations, reporting of serious AI incidents, and the activities of AI agents. The consultation also invites input on other areas where transparency gaps may exist.
The Centre’s submission identifies one such gap: non-human animals can be materially affected by AI systems, yet animal welfare is largely absent as a distinct category of impact within current AI governance frameworks.
AI systems are already being used in contexts that affect animals, including animal agriculture, veterinary medicine, biomedical research, wildlife monitoring and management, conservation, transportation, surveillance, and other forms of automated decision-making. These technologies can create benefits for animals, but errors, limitations, bias, inadequate oversight, or harmful applications can also produce welfare consequences at significant scale.
Transparency matters because what governments and institutions choose to document influences which impacts can later be assessed, studied, and governed. If animal-related impacts have no place within transparency or incident-reporting systems, significant harms may remain outside the evidence base available to policymakers.
Three recommendations for Canada
The Centre’s submission makes three practical recommendations:
1. Recognize non-human animals as potentially affected stakeholders in AI transparency.
Where AI systems are designed or deployed in contexts that materially affect animals, relevant transparency measures should account for foreseeable welfare impacts, system limitations, human oversight, validation, and mechanisms for reporting unexpected outcomes.
2. Capture significant animal harms in AI risk and serious-incident reporting.
Serious or systemic animal-welfare impacts—including death or serious injury, prolonged pain or distress, large-scale welfare consequences, or harmful automated decisions—should not fall outside reporting frameworks simply because the affected subjects are non-human.
3. Include animal-welfare expertise when transparency frameworks are designed.
Developing meaningful standards, risk classifications, incident categories, and reporting guidance requires the expertise needed to recognize the impacts being assessed. Depending on the context, this may include expertise in animal-welfare science, veterinary medicine, animal ethics, wildlife science, animal-protection policy, and related fields.
Closing the stakeholder gap
This submission does not propose a separate AI regulatory system for animals. Instead, it asks a more fundamental question: can Canada’s emerging AI governance infrastructure recognize material impacts when the affected subject is a non-human animal?
As Canada continues to develop its approach to responsible AI, the Centre for Animal-Aligned AI will continue examining where animal interests are represented, where they remain absent, and how those gaps can be addressed through practical policy and governance mechanisms.