Ottawa’s AI Transparency Consultation Needs Its Own Accountability Ledger

Ottawa wants Canadians to trust a new AI transparency process. Start with the federal government’s own AI ledger.

Innovation, Science and Economic Development Canada announced on July 23 that Ottawa is launching a public consultation on transparency for AI systems and AI-generated outputs. The consultation runs until September 23, 2026. That is a legitimate topic. It is also an immediate accountability test.

The government says the exercise centres on five issues: detecting AI-generated content, helping people know when they are interacting with AI, improving information about AI systems, tracking serious AI incidents, and tracking the activities and interactions of AI agents. If those standards are good enough to ask of the private sector, they are good enough for Ottawa’s own departments.

The consultation page invites Canadians to respond by survey or email. It also says submissions will be treated as public documents and may be posted online. Then comes the line that should make every privacy-minded taxpayer pause: depending on volume, AI tools may be used to process the feedback. If government uses AI to process a consultation about AI transparency, Canadians deserve to know which tool, which vendor, what data is retained, what human review applies, and whether any submission is summarized, categorized or filtered by machine.

The discussion paper points to a wider federal agenda, including Bill C-36 on privacy and consumer data, Bill C-34 on online safety obligations for social media services and some chatbots, targeted measures on deepfakes and election misinformation, a $50 million expansion of the Canadian AI Safety Institute, and a planned Trusted AI Certification program. That is a lot of regulatory machinery. The public should not have to guess how these pieces fit together.

Ottawa’s own record shows why a ledger is necessary. A 2026 FAccT/arXiv paper on Canada’s Federal AI Register analyzed 409 documented systems and found that 86% were deployed internally for efficiency. The authors argued the register can obscure human discretion, training and uncertainty management, producing “visibility without contestability.” In plain English: listing a system is not the same as letting citizens challenge how it affects them.

This is not an anti-AI argument. Conservatives should welcome useful technology that saves money, reduces backlog and improves service. But no government should be allowed to wrap automation in nice language while hiding the operating rules from the people affected by it.

Before Ottawa lectures Canadians on AI transparency, publish the federal AI accountability ledger: every departmental AI system, algorithmic impact assessment, vendor and contract, data source, retention rule, incident log, human-review requirement, appeal path, privacy assessment, consultation-processing tool and quarterly correction record.

Transparency cannot be a slogan attached to a black box. If the Liberals want Canadians to trust their AI agenda, the first rule is simple: show your own receipts.