By Dominic Naimool, Synomos
Introduction
In recent years, the surge of interest in artificial intelligence (AI) and automation has reignited discussions regarding the role of chatbots in modern business operations. From customer service to sales and support, chatbots have become a common sight across various industries, including the financial sector. Financial institutions, in particular, are exploring the potential benefits of integrating chatbots into their services to enhance customer experiences and streamline operations. With the ongoing modernization of banking systems in Canada, we are on the cusp of witnessing a new generation of chatbots in the financial sector. Unlike their predecessors, these advanced chatbots will not be limited to handling basic customer concerns. Instead, they are poised to engage in full-scale banking activities, including account management, transactions, investment advice, and more, in a far more sophisticated manner than articulating policies and providing information. Some leading jurisdictions have already given birth to chatbots of this vintage. Indeed, some financial institutions such as Raiffeisen Bank and Ally Bank have employed digital assistants with the capacity to perform tasks such as bill payments, balance inquiries and financial transfers on voice command. As the efficiency and optimization promised by artificial intelligence hastens the market towards the adoption of technology, executives must remain vigilant as the legal landscape evolves in turn. This article will discuss a development concerning chatbots and its potential impact on the financial sector.
Exploring Different Types of Chatbots
Before delving into the potential impacts, it’s crucial to understand the various types of chatbots commonly used in businesses today:
Menu or Button-Based Chatbots: These basic chatbots operate on a scripted menu where users select options to navigate through predefined pathways. They are useful for handling straightforward queries but may struggle with complex or nuanced interactions.
Rules-Based Chatbots: Building on menu-based functionality, rules-based chatbots use conditional logic to guide conversations. They work well for answering common questions but can be limited when faced with unpredictable queries.
AI-Powered Chatbots: The evolution of chatbot technology, AI-powered chatbots leverage artificial intelligence and natural language processing (NLP) to understand and respond to user input more dynamically. They can handle a wider range of queries, learn from interactions, and provide personalized experiences.
Voice Chatbots: These chatbots allow users to interact using voice commands, providing a more natural and convenient way to engage with the system. Voice chatbots are evolving with AI advancements and are increasingly capable of understanding spoken language.
Generative AI Chatbots: Representing the next frontier, generative AI chatbots are capable of generating new content based on learned patterns and interactions. They can create human-like responses and offer a more advanced level of interaction.
In general, beyond the irritation prompted by a chatbot’s pre-programmed “I don’t know” response, there is little complex risk presented by a rules-based chatbot or its kin. The risk profile changes greatly however as generative AI is introduced into the equation. The legal landscape surrounding chatbots received a significant development on February 14, 2024, when the British Columbia Civil Resolution Tribunal handed down a ruling that wrestled with the degree of liability that may be assigned for a representation made by a chatbot, in Moffatt v. Air Canada, 2024 BCCRT 149. Decisions of the Civil Resolution Tribunal are not binding on other decision-makers, but the ruling is persuasive and addresses a question financial institutions will face (see Isabelle St-Hilaire, “Case Comment: Lying Chatbot Makes Airline Liable: Negligent Misrepresentation in Moffatt v Air Canada” (2025) 58:2 UBC L Rev 591, 2025 CanLIIDocs 1963). The decision does not say what technology powered Air Canada’s chatbot, so its application to generative AI is an extension of the reasoning and not something the Tribunal decided.
Moffatt v. Air Canada, 2024 BCCRT 149
In November 2022, following the death of their grandmother, Jake Moffatt sought information regarding a bereavement fare from Vancouver to Toronto. Air Canada’s chatbot informed Moffatt that they could apply for the bereavement refund within 90 days of the issuing of the ticket. The words “bereavement fares” were a highlighted and underlined hyperlink to a separate Air Canada webpage titled “Bereavement travel” with additional information about Air Canada’s bereavement policy. Interestingly, the content of that page contradicted the chatbot and, as the Tribunal described it, stated that a bereavement fare could not be collected retroactively. Relying on the chatbot, Moffatt booked flights at the full fare and applied for the refund within the 90 days. When Moffatt applied, Air Canada refused and insisted that Moffatt had not followed the bereavement fare policy.
Moffatt pursued Air Canada for the difference between the fare that was paid and a bereavement fare, a difference of about $870. The matter was a small claims dispute decided on documentary evidence, with the parties unrepresented by counsel. Although the claim was not argued in these terms, the Tribunal framed the issue as one of negligent misrepresentation. To prove the tort, as set out in Queen v Cognos Inc, 1993 CanLII 146 (SCC), Moffatt had to show that Air Canada owed them a duty of care, its representation was untrue, inaccurate, or misleading, Air Canada made the representation negligently, Moffatt reasonably relied on it, and Moffatt’s reliance resulted in damages.
The Tribunal found that, due to their commercial relationship as service provider and consumer, Air Canada owed Moffatt a duty of care. Further, the Tribunal found that Air Canada did not take reasonable care to ensure its chatbot was accurate. While Air Canada argued Moffatt could find the correct information on another part of its website, the Tribunal could find no reason why the webpage should be considered inherently more trustworthy than the chatbot. In a fascinating submission, Air Canada argued that it is not liable for the actions of the chatbot by suggesting that the chatbot is a separate legal entity responsible for its own actions. The Tribunal was not convinced by this novel proposition and determined that the chatbot’s representations are considered representations on behalf of the company. Air Canada was found liable for the damages caused by the chatbot and therefore its misrepresentation. Generally, the damages for negligent misrepresentation entitle an individual to be put in the position that they would have been had the misrepresentation not occurred.
Liability Concerns: Financial institutions may be held responsible for the accuracy and reliability of information provided by chatbots. Moffatt itself involved a claim of about $870, but because damages for negligent misrepresentation follow the reliance, “minor” misrepresentations may have significant impacts. An example worth contemplating is that of a balance that is read out inaccurately due to the settlement of a cheque. Should the client make a major purchase or decision based on this incorrect information, it is unclear to what degree the financial institution would be liable. In addition, should the reasoning in Moffatt v. Air Canada be followed, the presence of correct information available at an ATM, with an in-branch teller or in the individual’s online banking account will arguably not shield the institution from liability. The contracts were not filed in evidence in Moffatt, so the Tribunal did not examine whether contractual terms could displace liability. Whether the terms of a member’s account agreement would change the analysis remains an open question. The liability for a financial institution would arguably extend beyond the limits of tortious misrepresentation and engage the Personal Information Protection and Electronic Documents Act (PIPEDA). Under PIPEDA, institutions bear the responsibility of keeping the personal information that they hold accurate to the degree necessary for the purposes for which it is to be used (Nammo v. TransUnion of Canada Inc., 2010 FC 1284).
Conclusion
The landscape of chatbots in financial institutions is rapidly evolving, driven by advancements in AI technology and changing legal considerations. As banking modernization takes centre stage in Canada, financial institutions must carefully consider the opportunities and challenges presented by the use of advanced chatbots. To ensure that this technology is used to its uttermost potential, that risks are sufficiently mitigated and the member trust relationship is respected, it is essential to ensure that chatbots are adequately trained and monitored.
Dominic Naimool
Dominic Naimool is a Barrister and Solicitor and the principal of Synomos Law.
General information only, not legal advice. Views are the author’s own.