There's a public consultation open right now that's easy to miss, and it's about something that affects anyone deploying AI.
On July 23, the federal government opened a consultation on AI transparency. It runs until September 23, and anyone in Canada can weigh in. The core question is simple: when AI is involved, can you actually tell what it is and what it's doing?
It's organized around five ideas worth knowing, because they're the same questions a careful organization already asks:
- Can you tell when you're dealing with AI and not a person?
- Can you spot content a machine generated or altered?
- Do you have clear information about what a system was built on, and what it can and can't do?
- Is there a way to record it when AI causes real harm?
- And for the newer "agentic" tools that take actions on your behalf: can anyone see what they're actually doing?
That last one is the part that keeps getting ahead of us. We've already seen what happens when a tool acts on its own and no one's watching the door.
The government's own framing is that "AI adoption moves at the speed of trust." That's true. But trust isn't a mood, it's built from exactly these things: knowing what a tool is, what it does, and who's accountable when it's wrong.
Consultations like this tend to be shaped by whoever shows up. The organizations actually using these tools day to day, not just building them, are the ones with the most useful feedback, and often the least likely to send it in. If AI is touching your work, it's worth twenty minutes to have a voice in the rules being written.
What this means for health organizations
The consultation's five themes (disclosure, labelling generated content, documentation, recording harm and overseeing agents) also make a practical checklist for your own AI policies, whether or not you took part.
