AI adoption and customer trust are inseparable. Without trust, financial institutions lack the consumer data and consent their AI needs to succeed. And without embedding trust at every step, AI initiatives falter. Success requires treating them as a single challenge, not two.
This insight emerged from a recent FT webcast in partnership with Gen, where leaders from Engine by Gen, American Express, Deutsche Bank, and HSBC discussed how financial institutions can deploy AI while building and maintaining customer trust. Together, they brought perspectives spanning consumer banking, wealth management, private banking, and fintech innovation.
Early in the discussion, one speaker laid out a framework that anchored the entire conversation: trust breaks down into three dimensions.
- Safety and security: Do you protect my data and money?
- Transparency and fairness: Do I understand the value exchange and know what to expect?
- The moment that matters: When something goes wrong, do you have my back?
These three dimensions ran through every aspect of the discussion—from how AI is deployed in banking, to the governance and accountability required, to who ultimately owns the customer relationship. Here's how these institutions are putting this framework into practice.
Upmarket AI adoption starts upstream
HSBC shared findings from its global research on trust and AI, conducted with over 10,000 private and wealth banking clients across 10 countries. The data reveals a clear pattern: 59% of respondents took action based on advice from a human advisor, while only 19% did so based on AI guidance. 57% still rely primarily on a human advisor.
For this affluent segment, this reliance rather than signaling a limitation, points to where AI creates immediate value. The practical opportunity lies upstream of the final decision: equipping advisors with faster, better information serves trust more effectively than automating decisions outright. For clients with significant wealth at stake, the realistic and optimal near-term goal is superior human judgment supported by AI.
This insight reveals a broader pattern. The more complex a client's situation, the less they want AI making decisions and the more they want human advisors. Wealth complexity, multiple asset classes, global considerations, family structures—these demand judgment, not automation. But here's where AI unlocks value: it can eliminate administrative burden. AI handles document processing, data reconciliation, and routine analysis, freeing advisors to spend time on what matters most—advice, listening, and what HSBC called "bedside manner." In other words, AI's real value isn't in replacing advisors. It's in empowering them to focus on the human relationship.
Automation with accountability
Speed should not come at the expense of governance. That was a core principle shared from Deutsche Bank's perspective. In KYC review, AI detects discrepancies between data sources instantly, but humans resolve them and own the judgment call that follows. This isn't automation versus human judgment. It's automation with accountability, where every decision is traceable to either a model or a person.
On the consumer side, the principle manifests similarly. Engine by Gen described how customers increasingly expect proactive insights like flagged credit changes and rate comparisons, delivered before they think to ask. Yet the underlying pattern remains the same: AI surfaces the information, humans make the decision. In both institutional and consumer contexts, AI's role is to notice. The human role is to decide.
There's a crucial tension here that Engine by Gen highlighted: customers appreciate speed when it comes with clarity, but they grow concerned when speed feels like a loss of control. The distinction matters: automation without added value erodes trust, while speed paired with meaningful insight strengthens it. When AI doesn't just process faster but delivers tangible context—identifying hidden opportunities, surfacing risks before they become problems, connecting data points humans would miss—it justifies the consent required to operate. That's when customers become willing to grant broader permissions: not because institutions ask, but because AI has demonstrated concrete value. So the goal isn't raw speed. It's speed that maintains or increases clarity about how decisions are being made and why, while genuinely improving outcomes for the customer. A faster decision that removes transparency erodes trust. So the goal isn't raw speed. It's speed that maintains or increases clarity about how decisions are being made and why.
Just because we can
As AI capabilities advance, two critical questions remain unresolved: whether we should deploy certain technologies simply because we can, and whether institutions can adequately defend against emerging security threats.
American Express raised two concerns. First, the discussion invoked the famous Jurassic Park line: you were so worried about whether you could that you never thought about whether you should. Second, voice and face cloning pose a specific identity risk that institutions must actively defend against. Together, these point to an unresolved question: how much autonomy should consumers be willing to grant, and who's responsible when agents fail?
Owning the relationship in a fragmented world
As customers increasingly research and transact on platforms banks don't control—16% of financial product searches now begin on LLMs—institutions face a fundamental shift. The customer relationship is no longer confined to the bank's channels. How do you maintain ownership when you can't control every touchpoint? The panel offered several perspectives.
American Express highlighted the risk that aggregators could become the dominant interface, weakening brand trust. Deutsche Bank argued that trust doesn't belong to whoever has the best interface, but to the institution that delivers consistently across the full relationship over time. Engine by Gen pointed out that trying to control every channel is impossible—the question is how to stay relevant when customers are distributed across platforms you don't own.
HSBC offered a framework that reconciled these perspectives: be the quarterback. Customers will encounter AI tools regardless of any single bank's efforts. The goal isn't to monopolize every touchpoint. It's to be the one they turn to when complexity strikes. The idea is to demonstrate to customers that “you've got my back no matter what the complexity or the issue is.” That's the trusted voice that matters.
Trust is the competitive advantage
As the discussion closed, one thing was clear: trust is evolving from a compliance and security concern into a strategic business asset. The winning institutions won't be those that automate fastest. They're the ones that understand what the discussion demonstrated: AI and human judgment aren't in competition.
When embedded responsibly, AI makes human judgment sharper, faster, and more trustworthy. Each institution will navigate this shift differently, shaped by their business model and client base. But the principle remains constant: the banks that earn trust are those that prove, consistently and transparently, that technology serves human judgment, not replaces it.
Trust is also evolving in another direction. Increasingly, customers care about the societal impact of AI itself. How do institutions address the energy demands of data centers? What's the impact on workforce evolution and job displacement? What's the environmental footprint? These aren't peripheral concerns—they're becoming central to how customers evaluate institutional trustworthiness. The winning banks will be those that address not just financial decisions, but the broader impact of the technologies they deploy.
Watch the full Financial Times webcast recording here.