Thanks very much for asking, Mr. Louis. I appreciate the question very much.
Let me give you a quick insight. I actually spoke on this with the committee in November, so it's quite nice to revisit it. Canadian election law has, over time, restricted how much can be done by so-called third parties to participate in our elections. We're talking about advocacy groups, unions or business associations advertising and being active during political campaigns.
The reason we've done this is that we've come to believe that political parties should be the principal vehicles through which our elections are contested. The longest ballot committee is actually an interesting example of a group that's acting like a third party, or maybe like a political party, but not putting itself into that regulatory framework. That's something for you to consider.
On the issue of AI and misinformation, I'll give you the following intuition, perhaps. We regulate the participation of third parties through two things. One is that we regulate how much they can spend, as a way of trying to limit how much they can speak. There are very strict spending limits on how much third parties can spend and thus engage in speech. The second thing we regulate is individuals. We make individual humans legally responsible for what third parties do.
The challenge with artificial intelligence is that it can bring the cost of communication down to zero. The spending mechanism limit on third parties isn't as effective when the cost of doing something is approaching zero. That's problem one.
Problem two is what you might call the agentic problem. We're not far away from people effectively creating non-human agents that will perpetuate speech during elections. Fake Twitter accounts are one example of this, but you can imagine an AI agent that's designed to create political advertisements and then put them out onto the web. It's not an individual putting them on; it's an autonomous agent. It's doing it at essentially zero cost, as it uses social media to spread itself.
What's needed in the legislative framework to address that is acknowledging that the human-centred spending limit approach to limiting third parties doesn't work as well, potentially, in an AI world.
I hope that's helpful to you.