Following the launch of HCLTech’s latest research, Future of Payments: AI everywhere. Trust nowhere?, Baskaran Subramaniam, Executive Vice President - Financial Services (Americas) at HCLTech, discussed the industry’s rapid shift from AI pilots to production at scale with tangible value. With virtually all payments leaders already using AI somewhere in their operations, the moment has clearly turned from if to how.
Why AI and where first?
AI is leading the agenda for payments organizations because of the “speed, efficiency and the experience it creates for customers,” and AI’s “ability to identify fraud and detect abnormal payments” in real time. Today, that spans location-aware card checks to instant decisioning at the point of payment.
That urgency mirrors the data: 99% of payment leaders already apply AI in some facet of operations or UX, with fraud detection one of the most common uses.
Friction versus risk: Finding the balance
Asked about the trade-off between smooth experiences and safety, Subramaniam explained that truly zero-friction doesn’t exist. “When you try to make anything less friction, risk can increase and regulators will be concerned,” he said. Organizations shouldn’t use AI to remove control, but to target it and “identify risk and abnormal payments much faster,” so most customers move seamlessly while risky patterns trigger smart checks.
Industry peers agree. 82% of executives say AI is the only viable way to balance a smooth customer experience with fraud prevention. Still, 91% are concerned about applying AI in payments, underscoring the governance bar that must be met.
Do risks vary by region?
On regional differences, Subramaniam framed today’s concerns as “early days” dynamics that will ease as models mature. The research highlights that executives in the US show higher concern about AI use (50%) than Europe (41%), with the UK & Ireland in between (45%). Many organizations also lack formal guardrails, with 47% reporting no AI policies, fueling anxiety around transparency and control.
Are AI fraud tools effective, yet?
Reacting to findings that 60% of leaders view current AI fraud tools as ineffective, Subramaniam called it a classic maturity curve: early models can misclassify and miss edge cases, but “like any other technology…the models will become more accurate.” The trajectory in the market points the same way: agentic capabilities are advancing, but confidence will hinge on better data, clearer policies and continuous tuning.
How to adopt AI safely and at scale
On practical adoption, Subramaniam advised starting small: “take the simplest process… pilot it,” prove value with minimal impact, then scale to more complex workflows and decisioning engines. That staged approach aligns with the structural reality in many institutions: only 20% have fully modernized, cloud-native, real-time data systems; most are still mid-migration, which is not ideal for industrializing AI. Meanwhile, customer pressure is rising: 87% of leaders fear losing customers without instant capabilities.
A clear and challenging path forward
The path forward is clear and challenging. Payments organizations are betting big on autonomy and intelligent decisioning, with 52% expecting to operate as autonomous financial services firms within 24 months, yet ambition still outpaces readiness. The winners will pair Subramaniam’s “pilot, prove and progressively scale” playbook with strong AI governance, resilient data platforms and transparent controls that earn trust. This will enable payment organizations to adopt AI that won’t just speed transactions, but quietly orchestrate smarter, safer and truly modern payment experiences for everyone.
