White warns AI tools are confidently misleading borrowers before they call a broker
For Canadian mortgage borrowers, the problem with artificial intelligence may not be what it doesn’t know — it’s how certain it sounds about what it gets wrong.
That is the core warning from Joe White, founder and chief executive of the Real Estate and Mortgage Institute of Canada (REMIC) in Toronto, who sees AI chatbots dispensing mortgage guidance with misplaced confidence as the profession’s next client management challenge.
“AI has a wonderful ability to give you an answer that sounds incredibly confident, but in many cases it’s the fine details that it gets wrong because it’s trained on everything out there on the internet — accurate information as well as inaccurate information,” White told Canadian Mortgage Professional.
The misinformation gap
White recently raised the issue directly with an Ontario Member of Provincial Parliament (MPP), having first tested it himself by prompting ChatGPT to act as a mortgage broker — an instruction the chatbot followed without hesitation, immediately soliciting the user’s financial information.
“Somebody goes online this morning and starts typing in and has a 15-minute conversation with Claude or ChatGPT or DeepSeek, and they walk out of there thinking, now I know what I have to do, and they don’t know that they could be wrong,” White said.
His proposed remedy is what he calls “gating” AI — restricting tools to curated, verified data sources rather than the open internet.
Without that guardrail, he argues, systems designed to give users responses they find satisfying become a liability in a high-stakes financial environment.
“I think it goes back to the entire open AI model — their goal is to have an AI that gives you information that you find valuable and helpful, but it’s wanting to please you. Whereas a broker’s job is to give you information that’s grounded in reality,” he explained.
The case for licensed expertise
White’s argument centres on accountability structures that AI cannot replicate. Licensed brokers carry errors and omissions insurance and are subject to provincial oversight, a layer of consumer protection that no algorithm can provide.
The Office of the Superintendent of Financial Institutions (OSFI) identifying AI as top risks to Canadian financial institutions underscores how far the formal regulatory framework still has to travel to match the pace of adoption.
“The regulators don’t have the ability to regulate AI like they do individuals,” White said.
As a practical measure, White recommends brokers ask every new client at intake whether they have consulted AI and what it told them.
Correcting AI misinformation, he argues, is a form of professional value that no chatbot can replicate.
Eitan Pinsky, owner of Pinsky Mortgages in Vancouver, previously told CMP that brokers who understand the limits of digital tools and invest in human expertise are best positioned for the AI transition.
White’s position is consistent: the broker’s edge lies not in competing with AI, but in knowing precisely where it falls short.
“AI, although it has its shortcomings, it’s here to stay and you have to adopt it while others are adopting it. But you have to do it in a very intelligent, meaningful, intentional way,” White noted.
Make sure to get all the latest news to your inbox on Canada’s mortgage and housing markets by signing up for our free daily newsletter here.
