SaaScada urges UK banks to move faster on AI deployment

UK banks risk falling behind fintechs and digital challengers on artificial intelligence adoption, according to Steve Round, co-founder and president of SaaScada, who cited research showing that only slightly more than half of UK banks have deployed AI. Round said the gap has market relevance because AI is already being used by more agile firms to improve customer insight, compliance processes and operating efficiency.

Round argued that many banks are treating regulatory uncertainty as a reason to delay substantive deployment. He said the Financial Conduct Authority’s outcomes-led approach has created ambiguity for firms, but should not be read as a block on adoption.

The debate comes as UK policymakers consider how financial services should use AI. MPs have criticised the FCA’s “wait and see” stance on AI in financial services, according to the Treasury Committee. Round said bank boards are also looking at the European Union’s AI Act for signs of how future obligations may develop, including more detailed reporting for AI systems and tighter governance for higher-risk uses such as credit scoring and AI-assisted investment activity.

Operational use cases

Round said many banks have concentrated on limited pilots or customer-facing generative AI tools, including chatbots, because those projects appear easier to demonstrate and carry less perceived risk. In his view, those applications often have less effect on banks’ core economics than systems that address internal processes.

He identified back-office and risk functions as the main areas where banks could gain from AI. The examples he cited include reconciliation, exception handling, interest accrual, payments monitoring, investigations, financial crime case assessment, batch processing, credit review and risk analytics.

In anti-money laundering work, Round said AI can help rank cases, reduce false positives and allow investigators to spend more time on higher-risk files when it is embedded with appropriate oversight. Such systems use models to process large volumes of alerts and supporting data, then flag items for review according to rules, patterns or learned indicators. Human review remains central where banks need to evidence decisions and meet regulatory obligations.

Data and governance

Round said banks cannot expand AI safely unless they address data quality, system fragmentation and oversight. He warned that weak data, inherited bias and insufficient controls could create problems at scale, especially once regulators demand clearer evidence of how models use information and produce outcomes.

He linked future AI controls to existing regulatory expectations, including the FCA’s Consumer Duty, saying banks should focus on accountability, control and evidencing decisions throughout their systems. That means institutions need to show why a decision was made and how a model reached it, according to Round.

For banks seeking wider deployment, Round called for modern core banking infrastructure, standardised processes that reduce exceptions, and governance that treats AI models as controlled components. He also said banks may need to work with outside partners, and in some cases industry peers, to share data, expertise and technology capabilities.

Round’s central argument is that UK banks should start with AI applications that cut operational strain and improve consistency, while building auditability and security into the design. He said waiting for more prescriptive rules risks leaving incumbents further behind firms that are already integrating AI into products and internal operations.

This story draws on original reporting from Finextra Research.