Fraud detection yes, loan approvals no: consumers draw the line on AI in banking
Hamburg, November 17, 2026
- 44 percent trust AI more than a bank employee when it comes to identifying suspicious transactions
- When applying for a loan, however, 46 percent prefer a human advisor, while only 24 percent would trust AI to make the decision
- 80 percent want financial institutions to disclose their AI policies
Whether people embrace or reject artificial intelligence in banking depends largely on how it is used. This is one of the key findings of the EOS Banking AI Barometer, a recent representative survey of consumers across Europe. The results show that trust in AI is not unconditional but depends on the specific use case.
“Forty-four percent trust AI more than a bank employee when it comes to identifying suspicious transactions,” says Janusch Skubatz, Chief Information Security Officer of the EOS Group. According to the study, no other use case achieved a comparable level of trust. “AI has proven its strength in fraud prevention. It can analyze vast amounts of data and identify patterns. In other areas, these capabilities are less obvious to consumers,” explains the information security expert.
This is particularly evident in lending decisions. Only 24 percent of respondents would prefer AI to make the decision when applying for a loan. Nearly twice as many (46 percent) would rather place their trust in a human advisor.
The study offers a possible explanation for these reservations about AI-driven lending decisions. Many respondents have specific concerns about AI’s ability to exercise sound judgment. “Forty-six percent of consumers cite the risk of mistakes or incorrect decisions as the biggest concern regarding AI in banking customer service, while 35 percent are worried about data privacy. This clearly shows that trust in AI does not emerge automatically,” says Skubatz. “Anyone looking to deploy AI successfully must do more than develop powerful systems. They must establish an AI Act-compliant framework and be transparent about how risks are managed.”
Consumers who see no risks at all are clearly in the minority, accounting for just 5 percent of respondents.