Bank customers are increasingly using generative AI tools to research financial products but remain reluctant to share personal data with them or allow them to act on their behalf, according to research published by the Deloitte Center for Financial Services last week.
Seventy-two percent of consumers said they are concerned about sharing information about their financial situation with gen AI tools, and 64% said they worry the tools may carry hidden biases toward certain banks or products. Forty-six percent said they trust the accuracy of banking recommendations from the tools.
Trust remained highest in banks’ own channels. Seventy-nine percent of respondents said they trust their primary bank’s website and materials, compared with 67% for traditional search engines, 62% for independent comparison websites, 49% for gen AI tools and 28% for social media or online forums.
Respondents drew the sharpest line at agentic AI. Eighty-three percent said they would feel anxious about an AI agent taking financial actions without their approval, 77% said such actions would make them feel less in control of their finances, and 66% said the risk of AI agents making errors outweighs the potential benefits. More than 40% cited costly errors, privacy and a lack of human oversight as concerns.
Among respondents who currently use gen AI, 58% have used it to research banking products, compared with 84% who have used search engines for that purpose. Sixty-five percent of Gen Z and millennial gen AI users have researched banking products with the tools, versus 55% of Gen X and 45% of baby boomers. Thirty-four percent of boomers said they trust gen AI recommendations, compared with 55% of millennials.
Trust tracked closely with experience. Among the roughly 1,700 respondents who have used gen AI tools, 60% said they trust the tools’ accuracy, versus 19% of those who have not.
Fifty-eight percent said gen AI saves them significant time, and 43% said it provides more useful information than banks’ websites or apps.
Deloitte recommended that banks track their presence in AI-generated shortlists, make product data machine-readable, require customer approval before AI tools act, and disclose the data and assumptions behind recommendations.
