Artificial intelligence (AI) is captivating the entire financial services industry, particularly for its ability to improve advisor productivity. But in an environment driven by risk management, the dangers associated with AI are a major concern.
In a February interview conducted by the Insurance Portal on the technological tools used by advisors, Céline Paret, President of Monarque Conseil, demonstrated the incredible potential of AI by providing the author of this article with the verbatim transcript, including a summary, of the interview she conducted using the Fireflies.ai application. This technology transcribes conversations conducted by phone or computer. While several applications offer this capability, the results provided by Fireflies.ai were astounding.
This technology, or similar tools like Google's PinPoint, is used by journalists at the Insurance Portal. The roles of advisor and journalist overlap in several respects when it comes to artificial intelligence, particularly regarding professional ethics.
Journalists must inform their interviewees that the information will be recorded. They must also protect their anonymous sources, which includes any information or data that could identify them. The results of their research must therefore be kept in a secure location.
Above all, journalists never take what they are told for granted and systematically cross-check the information they gather with reliable sources. There's a joke circulating in journalism schools: "If your mother tells you she loves you, check with another credible source." This is precisely what any advisor should do when consulting an AI application.
“The best way to achieve this is to ask the AI to cite its sources and obtain the hyperlinks to verify their accuracy,” says Emeline Manson, founder of CY-clic, an expert in fraud prevention and cybersecurity.
Verify the sources
It is therefore essential to systematically verify the information that AI provides, as this technology is known to sometimes deliver inaccurate or context-inappropriate information. It can even “hallucinate,” convincingly generating a completely false answer.
“AI remains general because it learns from the responses of other users,” adds Manson. “It can’t guess anything and ignores what it doesn’t know, unlike humans, who make quick connections in finance, investments, and insurance.” “
The expert, who also co-authored a recent guide in French on the use of AI in the financial services industry IA générative : mode d’emploi pour conseillers financiers avertis (Generative AI: A User’s Guide for Savvy Financial Advisors) with Virage Coaching and the business law firm Bernier Beaudry, adds that AI remains focused on the present moment. It struggles with planning and risk management. Advisors, on the other hand, have the ability to interpret more deeply and ask the right questions in cutting-edge fields.
However, AI is a tool that can dramatically boost productivity. “Many advisors would like to improve their efficiency, and from that perspective, AI seems like magic. But we must be wary of it,” says Willie Savard, founder of TChat N Sign, a technology provider for the financial services industry.
“AI should be seen as a human that needs constant training…If AI makes mistakes, the advisor is responsible,” he underlines.
During a panel on AI at the Congrès de l’assurance de personnes (Life Insurance Congress), an event organized by the Insurance Journal Publishing Group in November 2025, Savard stated that “it’s very difficult to find uses that comply with the law.”
Asked to elaborate on the subject in an interview, he believes that, generally speaking, advisors are not particularly tech-savvy, meaning they need help to meet compliance requirements. “In today’s technological environment, it’s quite simple: advisors shouldn’t use AI, whether free or paid, without asking themselves serious questions. Especially with free AI: they should stay away from it. For me, that’s a firm no.”
“People buy a $30-a-month package from ChatGPT or Claude and think they’re compliant. Nothing could be further from the truth,” comments Savard, who insists that professional ethics and the law prohibit sharing sensitive data with a third party without obtaining the consent of the person concerned. And AI is a third party.
AI data storage is one of the main issues. Aside from the Canadian company Cohere, virtually all active players in Canada are American. They use server farms often based in the United States. And even if they are located in Canada, certain American laws can give US authorities the right to obtain data without warning.
“And we don’t know where this data goes,” Willie Savard continues. “Americans have a more permissive approach to personal information than we do. Their compliance differs from ours. In practical terms, the GPU [Graphics Processing Unit] of your AI must be based on a Canadian server, and even then, we’re in a gray area. Above all, the questions and documents submitted to the AI must be anonymized so that no one can identify the client, even just by their investment habits.”
A GPU is the core of the technology that powers the AI. Unlike traditional processors (or CPUs), a GPU uses parallel data processing to perform thousands of calculations simultaneously, an important feature for training neural networks used in AI.
In the current context, Savard believes that the vast majority of people working in the financial services industry don't have sufficient knowledge of AI before using it. "Many call themselves experts, but they're just early adopters," he contends. "When you work with an AI tool, you must constantly keep in mind the legal impact of every action you take."
For her part, Emeline Manson emphasizes that regulators such as Quebec’s the Autorité des marchés financiers (AMF) and the Chambre de l’assurance require that data be stored in Canada. "Forget ChatGPT," she says. "Regulations require a Privacy Impact Assessment (PIA). Having a paid license isn't enough. You need to work in a more secure environment." “
Moreover, Manson is surprised to see that people prefer ChatGPT to Copilot, even though the latter is integrated with the SharePoint document management collaboration platform, itself linked to the Microsoft 365 suite. Copilot can therefore be used, provided that the organization to which one belongs has a Microsoft license that integrates this AI tool.
She emphasizes the unknowns surrounding this technology. “I have a friend who asked ChatGPT to generate a self-description for a dating app,” says Manson. “The text was amazing. So amazing, in fact, that it was frightening, because the AI knew far too much about her! We don’t really know how or what it based its description on. AI is chilling.”
Real risks
Sophie Babeux, partner and business coach at Virage Coaching, acknowledges that AI is very useful. “The problem is that most advisors don’t fully understand the risks,” she says. “That’s why they must systematically document all their interactions with it.”
She gives a concrete example: “If AI was consulted to develop a strategy, it must be documented in the client’s file. If the client suffers a significant loss and there’s an investigation, the process that led to the advisor’s recommendation must be explained in detail. If there’s an audit, the advisor will have to disclose that they used AI. They can’t simply state that they accepted Copilot’s recommendations; they’ll have to demonstrate how they verified them and the nature of the links to the client’s information.”
Babeux points out that there’s a significant difference between planning a trip using AI and planning a retirement strategy. However, she emphasizes that this technology is effective in other aspects of advisors’ practice.
“It optimizes brainstorming, email writing, strategic planning, and business processes in general. Some advisors and firms are already doing this, and the results are often astounding. AI often saves many hours of work. You just have to be aware of your ethical obligations,” summarizes Sophie Babeux.
Maintaining control
Fabien Major is one of those advisors who integrates several technologies into his environment, including AI.
“Conquest, my main financial planning tool, is the most advanced in the country,” he says. “It includes AI assistance that analyzes clients’ files and offers several ideas and elements that we might have overlooked, particularly regarding tax credits.”
According to him, this tool “is excellent for finding blind spots…Conquest doesn’t perform planning as such, but it’s a good assistant,” he explains. “It mainly improves the quality of presentations.”
“It handles things incredibly well, especially for tables and charts. You submit an idea, even just a verbal one, and the application creates a document from scratch. I can add visual elements, like charts or my logo.”
Fabien Major recently changed his customer relationship management (CRM) platform. He chose Maximiser, which includes a tool, IQ Boost, that allows him to quickly prepare meetings or compliance checks based on client files. He also uses an AI module integrated into Teams, Reader.ai, which summarizes client meetings and suggests actions accordingly. He doesn’t use Copilot.
Major’s firm is partnered with iA Financial Group, which offers proprietary account management technology with an AI module that generates statistics and creates Excel spreadsheets, enabling quick database searches to, for example, determine which clients hold which positions.
He sometimes challenges Claude with portfolio combinations. For example, for a conservative client, he requests suggestions for exchange-traded funds (ETFs). “I then analyze the results with other software to test their relevance,” he explains. “I have a professional subscription, so I remain in a closed system, but I systematically redact client data from all my requests. This technology is quite advanced when it comes to tax matters. I still validate the results, though. When I ask a question, I always request the sources, for example, an interpretation bulletin from the Canada Revenue Agency (CRA).”
The advisor recounts that for a client whose ETF was losing money, he asked Claude for a strategy suggestion, who advised him to realize a capital loss in exchange for a gain elsewhere in the portfolio. The recommendation proved correct.
He systematically informs his clients when he has used AI and how he used it. He adds a disclaimer clause to any text that has undergone analysis using this technology, but he also includes the sources. He believes that any self-respecting advisor must verify the AI compliance of the firm they are working with.
“We should never relinquish control to AI,” he continues. “We might get there in a decade, but the technology isn’t ready yet. Elon Musk has been trying to make his Tesla drive itself for 15 years, and it still sends him into dead ends. For me, AI is a tool like any other, but not a miracle cure. It simplifies some things and complicates others.”
Fabien Major points out that AI relies on past data. However, in the financial world, everything is constantly changing. You don’t progress by using old information. “Tax laws, financial regulations, geopolitics, society in general evolve according to emotions, feelings, and human behavior. This is not the channel of AI,” he concludes.
