Bank of Montreal’s CEO, Darryl White, raised concerns about the trustworthiness of artificial intelligence agents when it comes to handling personal banking data. Speaking at the Toronto Global Forum, White highlighted the growing gap between AI advancements and regulatory safeguards in the industry. This sentiment follows recent remarks by RBC’s CEO about the potential of restricting Meta’s popular Muse AI agent from accessing client information within the bank.
White emphasized that the ease and comfort of e-commerce have led individuals to freely share personal information with AI agents. He cautioned against providing sensitive data such as banking credentials, credit card details, social insurance numbers, or PINs, as these are not adequately protected. White stressed the importance of educating people on the risks associated with sharing such information.
In recent years, major Canadian banks have heavily invested in artificial intelligence technology. White noted that BMO’s insurance division utilizes AI to expedite underwriting decisions, reducing the process from 28 days to a mere 10 seconds. This rapid adoption of AI technology underscores the need for heightened awareness and caution regarding data security.
With client authorization, AI agents like Meta’s Muse and OpenAI’s ChatGPT can access and monitor bank accounts and credit cards to provide financial insights and spending advice. However, the need for human approval remains essential for making any transactions. RBC’s CEO, Dave McKay, mentioned that the bank is contemplating restricting Muse’s access to client data, balancing consumer needs with privacy and security concerns.
In light of these developments, leading financial institutions like JPMorgan Chase & Co. and Charles Schwab have taken proactive measures by either blocking or limiting external AI agents from accessing their clients’ data. This cautious approach reflects the industry’s growing understanding of the importance of safeguarding sensitive financial information in the era of advancing AI technologies.