Search
Szőke üzletasszony szürke blézerben áll egy modern épület előtt, mosolyogva nézi okostelefonját.

EOS Banking AI Barometer

What European consumers expect from AI in banking communication

Download whitepaper

Trust, expectations, and barriers: what consumers think about the use of AI in banks

icon red knowledge

Key findings of the EOS Banking AI Barometer

  • 46% of European consumers confirm that communication with their bank has improved through AI over the past three years, while only 11% perceive it as having worsened.
  • When it comes to detecting fraud, 44% of respondents trust AI more than a bank employee, while for lending decisions it is only 24%.
  • People who have already experienced payment difficulties are more open to AI in customer service.
  • 60% of 18- to 34-year-olds value AI’s round-the-clock availability, compared with just 47% of 51- to 65-year-olds.

Where the line is drawn: where consumers trust AI, and where they don’t.

Artificial intelligence is noticeably changing how consumers communicate with their financial institution. The EOS Banking AI Barometer examines how consumers across Europe rate the use of AI in banks today – where trust already exists, where reservations remain, and what expectations are placed on AI services.

For the study, consumers across Europe were surveyed on their experiences and attitudes toward AI-powered banking services. The results paint a nuanced picture: AI is not accepted or rejected across the board, but is always judged in the context of its concrete benefit. Overall, however, AI has already proven its worth in banking.

What do consumers expect from artificial intelligence in banks’ customer service today?

46% of European consumers say that communication with their bank has improved through AI, while only 11% perceive it as having worsened. Nearly one in three respondents (29%) would even prefer to talk to AI rather than a human on sensitive financial topics, yet 60% still miss the human touch in doing so.

Acceptance clearly depends on the benefit: 54% value AI mainly as help with finding information, while 51% each appreciate faster communication and round-the-clock availability.

When it comes to payment difficulties, or when an overdraft facility has already been fully exhausted, many bank customers want significantly more active support from AI. Approval is especially high for AI-powered warnings about an impending account overdraft: 61% of respondents would welcome such alerts. What else do consumers expect from AI in banking?

Egy mobiltelefont tartó nő mesterséges intelligenciával beszélget.

Perceptions of AI in everyday banking

Top 2 box on a scale from 5 “strongly agree” to 1 “strongly disagree”

Bank customers expect AI in customer service to deliver speed and availability above all; for more complex, judgment-based tasks, approval remains considerably lower. How do respondents rate the following statements about the use of AI in banks’ customer communication, broken down by age and prior experience with payment difficulties? 

Answer Total 18-34 years 35-50 years 51-65 years Payment difficulties No payment difficulties
AI makes communication faster. 51% 59% 51% 48% 59% 49%
AI makes communication friendlier. 37% 45% 38% 33% 46% 34%
AI offers better service because it's available 24/7. 51% 60% 50% 47% 57% 48%
AI helps me find information more easily. 54% 60% 55% 51% 61% 52%
AI provides more precise answers. 36% 43% 37% 33% 46% 32%
AI understands my concerns well. 35% 42% 35% 32% 44% 32%
AI can solve my problems effectively. 35% 45% 34% 32% 44% 31%
For sensitive topics, I prefer talking to AI rather than a human. 29% 38% 30% 24% 39% 25%
I miss the human touch in communication with AI. 60% 53% 59% 64% 61% 60%
AI scares me (e.g. because of data protection). 38% 42% 37% 37% 42% 36%
AI helps me keep track of my finances. 34% 41% 36% 29% 44% 30%
Marwin Ramcke, az EOS Csoport igazgatótanácsának elnökének portréja egy világos szobában, sötét blézerben

Our study shows a clear direction in the use of AI services: customers want AI that is not only reactive but also proactive. They are ready for smarter services; now the banks need to deliver.

Marwin Ramcke
CEO EOS Group

How much do consumers trust AI with lending decisions?

Trust in AI depends heavily on the use case. It is highest where AI mainly ensures security: 44% of respondents trust AI more than a bank employee when it comes to detecting fraud and suspicious transactions – the highest score of all scenarios surveyed. For lending decisions, the picture reverses: only 24% prefer an AI decision here, while 46% clearly trust the human advisor.

Skepticism, however, doesn’t automatically mean rejection: consumers tie their trust to clear conditions. For 77%, it is important that banks disclose when and how they use AI. The biggest concern remains AI’s susceptibility to error (46%), followed by data protection concerns (35%) and a lack of empathy (28%).

Egy fiatal nő áll a városi környezetben, kezében okostelefonnal, mosolyogva néz előre. A háttérben modern épületek és fák láthatók.

Trust in AI by use case

Top 2 box on a scale from 5 “significantly more trust in AI” to 1 “significantly less trust in AI”

Trust in AI varies greatly by use case: it is highest for fraud detection and lowest for lending decisions. How much do respondents trust the decisions of an AI compared with the decisions of a bank employee in the following scenarios? 

Answer Total 18-34 years 35-50 years 51-65 years Payment difficulties No payment difficulties
AI decides whether I get approved for a loan. 24% 29% 26% 20% 34% 19%
AI suggests suitable investments / investment recommendations to me. 31% 38% 31% 28% 39% 28%
AI identifies suspicious transactions / fraud attempts on my account. 44% 47% 43% 42% 49% 42%
AI suggests tailored products to me. 37% 44% 38% 32% 45% 34%
AI advises me in case of payment arrears / liquidity shortages. 32% 40% 31% 28% 41% 28%
AI can effectively manage my regular transactions. 32% 37% 32% 31% 40% 29%

Risks and concerns about using AI in customer support

Consumers’ biggest concern is AI’s susceptibility to error, followed by data protection concerns. Where do respondents see the greatest risks and concerns about the use of AI in banks’ customer support?

Answer Total 18-34 years 35-50 years 51-65 years Payment difficulties No payment difficulties
Data protection and security of my personal data 35% 39% 32% 36% 32% 37%
Lack of empathy 28% 26% 28% 28% 29% 28%
Fear of errors / wrong decisions by AI 46% 45% 47% 46% 44% 48%
Lack of transparency about how decisions are made 21% 19% 21% 22% 20% 21%
The feeling of not being treated as an individual 22% 20% 23% 21% 23% 21%
No help with individual concerns 25% 23% 24% 26% 27% 24%
Possible discrimination through AI algorithms 19% 18% 19% 18% 22% 17%
Loss of jobs at the bank due to AI 30% 32% 30% 30% 31% 31%
I don't see any critical points. 5% 3% 6% 6% 4% 5%

What support do people with payment difficulties want from AI?

Those affected mainly want proactive suggestions from AI. 30% of all respondents have already had trouble paying an instalment to their bank or have exceeded their overdraft facility, and 25% have already been in contact with a debt collection company. People with this experience are considerably more open to AI support offers than average: 61% of those affected would welcome AI suggesting that they pause an instalment, compared with 52% of those not affected.

The desire for human support nonetheless remains: 80% of those affected want to involve a human when clarifying payment difficulties, either exclusively or after an initial AI contact. However, 13% of those affected could also imagine communicating exclusively with AI, compared with just 6% of those not affected.

Egy mosolygó, göndör hajú fiatal nő egy kényelmes sárga pulóverben ül a kanapén a nappalijában, és az okostelefonját nézi, miközben egy papíralapú dokumentumot tart a kezében.

Concrete instalment plans

Those affected want AI to actively propose a concrete instalment plan when money is tight: 

0

percent
of respondents

“AI lowers the barrier to dealing with financial problems – people step in where empathy and individual solutions are needed.”
Daniela Moaca, Senior Manager IT Romania

Daniela Moaca, Senior Manager IT az EOS Romaniánál, portréfotó sötét garbóban.

Proactive advice

Those affected want AI to suggest pausing an instalment:

0

percent 
of respondents

How does the younger generation differ from the general population in how it uses AI?­

18- to 34-year-olds are more open to AI offerings than older age groups across almost every scenario: 60% of 18- to 34-year-olds value round-the-clock availability, compared with just 47% of 51- to 65-year-olds. Younger respondents also value the speed of communication considerably more often, at 59% (versus 48% among older respondents). The difference is clearest in actual chatbot use: among younger respondents it stands at 28%, more than double the 13% recorded for the older age group.

This openness to AI continues when it comes to payment difficulties: 14% of 18- to 34-year-olds would communicate exclusively with AI about payment difficulties, compared with 8% on average overall. Only 38% of this age group insist on an exclusively human point of contact – the lowest figure of all age groups.

Egy fiatal férfi fekszik a nappalijában egy kanapén, fejhallgatót visel, és mosolyogva nézi az okostelefonját, miközben zenét vagy podcastot hallgat.

Communication preference for payment difficulties by age group

Younger bank customers are more open than average to almost all AI applications, including sensitive topics such as payment difficulties. When it comes to clarifying payment difficulties or debts with the bank, who would respondents prefer to communicate with? 

Answer Total 18-34 years 35-50 years 51-65 years Payment difficulties No payment difficulties
Exclusively with a human 46% 38% 47% 50% 41.6% 48.6%
Preferably with a human, but AI as a first point of contact is fine 39% 42% 39% 38% 38.7% 39.6%
Preferably with AI, since the topic is uncomfortable for me 8% 14% 8% 6% 13% 6%
I don't mind (human or AI) 6% 7% 6% 7% 7% 6%
Egy illusztráció egy embert ábrázol, aki barátságosan átkarol egy robotot. Alatta az "AI-powered. Human-led." szöveg látható, amely az emberek és a mesterséges intelligencia közötti együttműködést szimbolizálja.

Conclusion: AI in banking as a complement, not a replacement

The findings of the EOS Banking AI Barometer show that the use of AI in banks is accepted when it offers a clear benefit, is communicated transparently, and complements rather than replaces access to human support. Banks that strike this balance lay the foundation for lasting trust among their customers.

About the study

Checklist Decision Icon Icon depicting a document or checklist with a checkmark and an 'X' mark, symbolizing task management, decisions, approvals, rejections, or evaluating options.

Institute

The representative survey was conducted online on behalf of EOS by the market research institute PureSpectrum in July 2026.

icon black website

Countries

11 countries: Bulgaria, Croatia, France, Germany, Greece, Poland, Portugal, Romania, Slovenia, Spain, Switzerland

icon black team

Methodology

Online survey (CAWI), 4,400 respondents (400 per country), consumers aged between 18 and 65 who are primarily or jointly responsible for managing their household’s finances, representative by gender and age.

Frequently asked questions

4,400 bank customers in 11 European countries, 400 per country, surveyed online in July 2026.

It depends on the task: for fraud detection, 44% trust AI more, while for lending decisions it’s only 24%.

13% of those affected would communicate exclusively with AI about payment difficulties, with no human contact at all. The desire for human support nonetheless remains dominant: 80% want to involve a human, whether from the start or after an initial AI contact.

Yes: 18- to 34-year-olds are considerably more open to almost all AI applications, for example when it comes to round-the-clock availability (60% vs. 47% for 51- to 65-year-olds).

The biggest concern is susceptibility to error (46%), followed by data protection (35%) and a lack of empathy (28%).

Questions about the study? Feel free to contact us.

Lara Flemming, Senior Vice President Corporate Communications & Marketing at EOS

Lara Flemming

Senior Vice President Corporate Communications & Marketing


[email protected]