As artificial intelligence continues reshaping the way Canada’s financial institutions operate, the Office of the Superintendent of Financial Institutions (OSFI) has made one thing clear: it has no plans to rush into regulating AI. Superintendent Peter Routledge (pictured right) told attendees at the Thomson Reuters conference that premature rule-making could destabilize rather than safeguard the system.
Routledge said OSFI’s instinct was not to jump in early with controls or restrictions.
“We could try and dream up how to regulate this extraordinary new technology, but I think if we did, we would probably be acting too early and subject ourselves to the risk of doing more harm than good," he said.
Instead, he signalled that OSFI is deliberately choosing curiosity over intervention – even if that means accepting some uncertainty.
By “standing back from it,” he said, OSFI accepts “the risk that [AI] could get applied the way that into unintended and unwanted risk concentrations,” but believes that allowing institutions room to experiment is essential “to enable our institutions to innovate with the tool.”
Routledge emphasized that boards and executives – not OSFI – must determine how to integrate AI responsibly into their operations.
“Honestly, these institutions with their senior execs and their boards have to sort that out, and we don’t want to interfere.”
While OSFI is avoiding early sector-wide regulation, it has been building its own internal capacity with artificial intelligence – a process Routledge described as cautious, sometimes uncomfortable, and ultimately necessary.
When public AI tools first became widely accessible, OSFI took a hard line.
“As AI public AI tools became widely available on the internet, we initially closed down all access to them because we didn’t want the private information we have to be in these public models, and that was necessary, painful, but necessary.”
Only after locking down external exposure did OSFI begin investing in its own controlled AI environment.
“What we’ve subsequently done is we’ve acquired internal models to help our supervisors and our regulators leverage this tool to be more productive, to make better decisions faster.”
OSFI is now piloting targeted AI use cases within supervisory and regulatory teams as it works to understand how the tools can safely enhance oversight.
“We have started a pretty bold program of experimentation with AI internally… we’re calling individual use cases for specific supervisory or regulatory teams and specific issues, and we’re trying to figure out how we’re going to use it to become more productive.”
Later in the session, Routledge addressed a question from Insurance Business on which elements of the federal budget will have the greatest impact on the insurance sector. While he noted that the most immediate changes are aimed at the banking system, he stressed that the implications will inevitably flow downstream to insurers.
“I’d give you two answers to that, and I wouldn’t say these two answers affect the insurance industry first or foremost; they are more banking-system related, but they’re so fundamental, they will [have an effect].”
His first point centred on Ottawa’s move toward legislating stablecoins – a signal, he said, that Canada is preparing for a genuine digital asset regulatory framework.
“It really signals, in my view anyway, Canada’s intent to have a digital asset regulatory framework, and digital assets in our system are going to start with a payment asset we’re going to call stablecoin.”
That shift, he added, could also pave the way for incumbents to explore tokenized deposits.
“That will probably also usher in innovation in incumbents around tokenized deposits – we’ll see. But [it’s a] reasonable thing to plan for, and that’s going to change the way money’s going to flow through the system. And insurers are going to have to be ready for that. The traditional flow of regulated deposits is going to change.”
Routledge’s second point was the budget’s extensive anti-money-laundering (AML) provisions – including structural changes that will deepen ties between OSFI, FINTRAC and the newly proposed financial crimes agency.
“The second is all the AML provisions, anti-money-laundering provisions that are in the act – the apparent start of a financial crimes agency, moving FINTRAC onto OSFI’s Financial Institution Supervisory Committee, which allows us to exchange information between OSFI and [FINTRAC] more easily.”
The legislation, he said, will also increase OSFI’s visibility into risks that already exist in the system.
“There’s legislation that is going to create another risk for OSFI – maybe it creates the wrong one. It’ll give us greater vision into an existing risk, and that’s going to have a consequence in terms of how we supervise.”
For financial institutions of all kinds, including insurers, it becomes clear that AML vigilance is no longer optional, he said.
“Money laundering is… an attack on your institution by a hostile entity, and institutions have to protect themselves from those hostile threats and attacks.”
