Canada’s AI Consultation Closes September 23
Innovation, Science and Economic Development Canada has been running a public consultation since July 23 called Enhancing trust in artificial intelligence through increased transparency. It closes on September 23. Five areas are on the table, three of them would land directly on ordinary businesses if anything comes of them, and the submissions will mostly come from organisations with government relations departments.
The five areas
| Area | If something came of it |
|---|---|
| Identifying AI-generated content | Labelling or provenance metadata on what you publish |
| Knowing you are talking to AI | Disclosure on chatbots and automated phone handling |
| Information about AI systems | Vendors publishing capabilities and limitations |
| Tracking serious incidents | A duty to report when an AI system causes harm |
| Tracking AI agent activity | Logging what agents did, and identifying them as agents |
On the first area, the consultation names visible disclosures, invisible watermarks and provenance metadata as possible approaches, and asks both when identification matters most and which mechanisms would work. On the second, it asks whether and in what circumstances people should be told they are dealing with a system rather than a person.
The third area is the one most favourable to buyers. Consistent information about what an AI system can and cannot do is precisely what is missing when a small business tries to evaluate a vendor, which is the gap behind AI washing and behind the difficulty of reading accuracy claims covered in the 54% legal AI benchmark.
What is not being decided
Being accurate about this matters, because the alarmed version of the story is easy to write and wrong. The discussion paper asks whether existing market practices and legal frameworks are already sufficient, or whether additional measures are warranted.
The measures on the table span a wide range: legislation or regulation at one end, but also government guidance, voluntary codes of conduct, technical standards, research initiatives, AI literacy programmes, and procurement requirements. Procurement is quietly one of the more consequential, since a rule that applies only to government suppliers still reshapes what vendors build.
ISED has said it will publish a What We Heard report after the consultation. No timeline commits Canada to any of this, and anyone telling you new AI rules arrive next year is guessing.
Three areas that would reach a small business
Chatbot disclosure. If you run a website chatbot or automated phone handling, a disclosure requirement is a small change and a real one. Most businesses running these have never written down whether callers are told. That is worth knowing regardless of what Ottawa decides.
Content labelling. If you publish marketing material, product photography or written content produced with AI, labelling obligations would change your process. Our note on AI watermarks covers how the technical side currently works, which is less settled than the policy discussion implies.
Agent activity tracking. The newest area and the one with the least established practice. If agents act in your systems, logging what they did is something you should already be doing for your own protection, as we argued in privileged access management. A requirement would formalise it rather than invent it.
How to respond before Wednesday
Two routes, both closing September 23. There is an anonymous online survey on the ISED consultation page, and you can send written comments to AIConsultations-ConsultationsIA@ised-isde.gc.ca.
One detail worth reading before you write anything. ISED states that submissions will be considered public documents and may be posted online, and asks that you not include personal or confidential information. Keep client names, contract details and anything commercially sensitive out of it.
If you are writing rather than taking the survey, the most useful thing a small business can contribute is specifics. Not a view on whether AI should be regulated, which the policy team has heard, but a paragraph on what a particular obligation would actually cost you to comply with, who in your business would do it, and what would happen to the tools you currently rely on. That kind of detail is scarce in consultation responses and disproportionately useful.
Why bother
Consultations like this are answered by companies with government relations functions and by advocacy organisations with a position to advance. The operating reality of a ten-person firm in Saskatoon is under-represented by default, not by design.
Whether one submission changes anything is genuinely uncertain, and it would be dishonest to claim otherwise. What is certain is that the positions on record now form the input to whatever comes next, and that the window for adding yours closes Wednesday. We covered the consultation when it was announced in our note from August, and the broader policy direction in what Bill C-36 would mean.
Frequently Asked Questions
What is the ISED AI transparency consultation?
A public consultation run by Innovation, Science and Economic Development Canada called Enhancing trust in artificial intelligence through increased transparency. It opened on July 23, 2026 and closes September 23, 2026, and supports Canada’s National Artificial Intelligence Strategy: AI for All. It asks for views on five areas where transparency issues arise and where ISED says further action may be warranted.
What are the five areas?
Detecting and identifying AI-generated content. Empowering individuals to know when they are interacting with an AI system. Improving the availability of consistent and understandable information about AI systems, including development, capabilities and limitations. Enabling the tracking of serious incidents related to AI systems. And advancing ways to better track the activity and interactions of AI agents.
Does this mean Canada is about to regulate AI?
Not necessarily, and the discussion paper is explicit that it asks whether existing market practices and legal frameworks are sufficient or whether additional measures are warranted. The range of possible outcomes runs from doing nothing through guidance, voluntary codes of conduct, technical standards and procurement requirements to legislation or regulation. Treating this as imminent rules would be overstating it. Treating it as irrelevant would be a mistake of the opposite kind.
How do I respond to the consultation?
Two routes. An anonymous online survey, or written comments by email to AIConsultations-ConsultationsIA@ised-isde.gc.ca. Both close September 23, 2026. One important detail: ISED states that submissions will be considered public documents and may be posted online, so do not include personal or confidential information, client names or anything commercially sensitive in what you send.
Is it worth a small business responding?
The consultations that shape rules tend to hear from large companies with government relations teams and from advocacy organisations, so the practical experience of a ten-person firm is under-represented by default. If any of the five areas would create real work for you, a few paragraphs describing why is more useful to a policy team than a general opinion. Whether it changes the outcome is uncertain. The cost is twenty minutes.
Know which rules would actually reach you
We map where AI touches your operations and tell you plainly which proposed obligations would create real work and which would not apply at all.
Related Articles
Billable Hours When AI Does Part of the Work
Only 9% of Canadian Small Businesses See AI Gains
Canadian AI Companies: What Counts and Why It Pays
Ajan leads the ChatGPT.ca team: 200+ custom GPT builds and automation projects for 50+ businesses across 20+ industries. Based in Markham, Ontario. PIPEDA-compliant solutions.