Ottawa Is Drafting AI Transparency Rules. Speak Now.

Innovation, Science and Economic Development Canada has a public consultation on AI transparency open until 23 September, part of the AI for All strategy launched in June. It asks how Canada should handle five specific things, and unlike most government consultations, all five land directly on how ordinary businesses already use AI. Almost nobody outside large firms and industry associations will respond, which is exactly why a thirty-minute reply from a small Canadian business is worth more than it sounds.

The five questions on the table

Rows two and five are the ones to read carefully if you run a small business. Disclosing AI interaction touches every company with a website chat assistant, an AI phone answering service, or automated email replies, which by now is an enormous number of perfectly ordinary firms. Agent monitoring matters because agents are moving from experiment into production faster than governance is following, and any recordkeeping duty lands on whoever deployed them rather than whoever built them.

Why your response counts more than you think

Consultation responses come overwhelmingly from large companies, industry associations, and advocacy organisations, all of whom employ someone whose job includes writing them. The lived reality of a twenty-person firm in Kitchener or Moncton rarely reaches the file at all. That is how rules end up technically sensible and practically awkward for smaller operators, not through malice but through absence. Describing what you actually do, and which proposed obligation would be genuinely hard for a business your size, is information nobody else in the process is providing.

Nothing is mandatory, and that is the point

This is a consultation about what rules should look like, not a rule. Your existing obligations are unchanged: PIPEDA federally, Quebec's Law 25 if you handle personal information there, and whatever governs your sector. We set out the current landscape in what actually regulates AI in Canada in 2026. The reason to engage now is timing. Once a framework is drafted the direction is largely settled and everyone is arguing about details. The consultation stage is the only point where the shape itself is genuinely open.

The work that is useful either way

Whatever emerges, every plausible version starts from the same place: knowing where AI touches your business. Write the list. What each use does, who owns it, and whether customers are told. That inventory is not speculative compliance work, because you need it for ordinary management too, and most small businesses do not have it, which is the gap behind using AI without managing it. An afternoon now beats a scramble later, and it produces a better list than one written against a deadline.

Disclosure is the one to get ahead of

Of the five areas, telling customers when they are dealing with AI is the one most likely to become a firm expectation, because it is the easiest to write, the easiest to check, and the one the public most clearly wants. It is also the cheapest to adopt voluntarily, and businesses that do it early tend to find customers respond well rather than badly, which is the argument in when to disclose AI to customers. Doing it before you are required to converts a future compliance task into a present trust advantage. That is a rare trade and it is available now.

Thirty minutes, before 23 September

ISED accepts responses through an online survey or by email, and you do not need legal drafting or a policy position. Plain description works: here is what we use AI for, here is how we tell customers, here is what would be difficult and why. The wider strategy this sits inside is covered in Canada's AI for All strategy. Check the ISED page for the current deadline and submission route before you write, then take half an hour. Rules get written either way; the only question is whether anyone in them sounds like you.

Frequently Asked Questions