The New Foundations of Trust in Finance: What Banks and Fintechs Need to Get Right
- Gustavo Lino, Head of Policy, Legal and Compliance at Cumbuca
- 21.09.2026 08:45 am #fintech
Financial data now moves between institutions more freely than ever, and customers are asked to share more of it while trusting decisions made by systems they can't fully see. Onboarding has grown more complicated and recommendations more automated, and customers have grown more cautious as a result. What decides whether trust holds up in this next phase is less about the technology and more about whether processes stay visible, whether customers keep real control over their own data, and whether an institution owns up when AI gets something wrong. Get it right and a bank keeps the customer. Get it wrong, and that customer is gone within a few taps.
What Pix gets right
Pix shows what happens when the first of those gets done well. Brazil's instant payment system overtook credit cards in online purchases last year, 42% to 41%, and is forecast to account for half of Brazil's e-commerce transactions by 2028. It got there because using it feels the same every time. There's no credit check and no cost, and it's built into nearly every banking app in the country. There is also a clear incentive for both merchants and customers: merchants receive Pix payments instantly without having to pay acquirer fees, and often pass some of those savings on by offering customers a 1-5% discount for paying by Pix.
That reliability, more than any clever feature, is what has drawn in customers who'd never used a formal financial product before. That matters more in a market like Brazil's, where switching between banks and fintechs already happens with a few taps. Open Finance builds on the same foundation, letting people share their data across banks, lenders and fintechs so they can reach credit and insurance many were previously locked out of.
Consent has to mean something
Reliability only carries an institution so far if customers can't see what happens after they hand over their data. The UK's Open Banking has processed more than a billion payments since it launched through the country's biggest banks, so people clearly will use connected finance once they trust it. Getting that trust right means consent has to mean something in practice: who can see the data, what they're allowed to do with it, how long they keep it, and how to switch it off without filling in three separate forms.
Concretely, that means a dashboard showing every active data-sharing permission in one place, with a record of when each one was last used and by whom. Cutting off access should take one tap and take effect immediately. Control comes from revocation and visibility, not from forcing customers to re-authorise a connection they still want.. Most Open Finance products still fall short on at least one of these points. Feel confused about any of that and a customer just leaves. Switching banks now takes minutes.
AI raises the stakes
AI adds a further layer of pressure. Bank of America's Erica has surpassed 3 billion customer interactions since launching in 2018, which shows how central AI assistants have become to how millions of people interact with their bank. The FCA has said, through its work on AI and the Consumer Duty, that firms remain accountable for how AI is used and that senior managers carry that accountability personally. The regulator is looking further ahead too.
A review published in July 2026 and led by the FCA's Sheldon Mills found that a fifth of UK adults are already comfortable letting AI make financial decisions for them, and that roughly a quarter would trust general-purpose tools such as ChatGPT or Gemini for financial advice, often unaware that these platforms offer no formal route to recourse if that advice turns out to be wrong. Mills warned that AI can amplify “misleading or hallucinatory advice” as much as it can amplify efficiency, and recommended tighter oversight as more financial decisions get delegated to AI systems. In practice, customers should be able to tell when a recommendation came from a model rather than a person, and should have somewhere to go if that recommendation turns out to be wrong.
The commercial cost of losing trust
The commercial argument for taking this seriously is straightforward. A Harvard Business Review study tracking 360 privacy-related company announcements over 14 years found that brands with strong privacy reputations saw a 12.31% increase in customer patronage, and that the effect was strongest among companies that had already survived a data breach. Trust pays for itself, and it compounds the longer a company holds onto it. Once it breaks in a market this easy to leave, it rarely repairs itself. That's a bigger opportunity for fintechs than for established banks, since a newer challenger has more room to build the kind of reputation incumbents already have.
Where banks and fintechs should start
None of this needs new legislation to start, and much of it sits closer to product design than to policy. Real-time status tracking in onboarding is one place to begin. Making consent as easy to withdraw as it is to give is another. Add an audit trail behind every automated decision a customer might question, and a bank has covered the basics its customers are already asking for. Getting there means compliance and product teams working from the same brief rather than handing requirements back and forth after the fact. The ones that manage that will be the ones customers still trust once switching gets even easier than it already is, and once regulators start asking harder questions about how AI decisions get made.
