Canada’s federal government has initiated a public consultation on artificial intelligence transparency, seeking input on how AI systems should disclose their identity, capabilities, and limitations. The initiative, announced by AI Minister Evan Solomon, aims to shape future policies and regulations around chatbot disclosures, AI-generated content labeling, incident reporting, and accountability for AI agents.
Key Areas of the Consultation
Consultation documents focus on five main areas: content labels for AI-generated material, chatbot disclosure requirements, system documentation, incident reporting protocols, and tracking of AI agents. Minister Solomon emphasized the importance of transparency as AI becomes more integrated into daily life, asking Canadians whether they can identify when they are interacting with machines.
AI Risks and Incident Reporting
The government is examining how companies should report serious incidents linked to AI systems, including security breaches like prompt injection attacks. Officials are also considering standard documents explaining intended uses, known risks, and safety controls. The consultation cites examples of AI-generated legal citations that have led to errors in Canadian courts since 2024.
Other topics include whether developers should disclose training data sources, including copyright-protected material, and technical measures such as digital watermarks. The government is also exploring mandatory incident reporting for security failures, financial losses, or unauthorized system actions.
AI Agents and Accountability
Special attention is given to AI agents capable of booking meetings, sending emails, making purchases, and completing forms. The government wants to clarify when agents must identify themselves and how users can track automated actions. An example illustrates responsibility issues when an agent misreads instructions and orders 12 packages instead of one.
Officials are studying how AI agents interact with other automated systems, considering rules on permissions, human oversight, transaction records, and identity disclosure.
Broader Policy Context
This consultation aligns with other Canadian measures: a digital safety bill requiring platforms to label synthetic content, a privacy bill mandating explanations for automated decisions affecting employment or credit, and a law criminalizing non-consensual sexual deepfakes. Proposed rules also address AI-generated election misinformation. The consultation supports Canada’s ‘AI for All’ strategy launched June 4, 2026.
Canadians can submit feedback until September 23, 2026. The government plans to publish a ‘What We Heard’ report after the consultation closes.