Financially vulnerable consumers are finding it increasingly difficult to access support from financial services providers, and poorly implemented artificial intelligence (AI) may be making the problem worse. New research from ArvatoConnect, based on a survey of 1,000 senior decision-makers across UK banks, insurers, fintechs, building societies and credit providers, highlights a growing gap between firms’ AI ambitions and the needs of vulnerable customers.
The study reveals that 74% of financially vulnerable people have felt like giving up while trying to get help from their bank, insurer or other financial provider, while 26% abandoned their attempt altogether. At the same time, many industry leaders acknowledge that AI-driven customer service systems may be contributing to these difficulties. More than three-quarters (77%) of finance leaders believe their organisation’s AI strategy poses a risk to vulnerable customers, with 28% describing that risk as high.
Despite these concerns, AI adoption continues to accelerate. Nearly nine in ten (88%) financial services organisations have increased their use of AI in customer-facing operations over the past year, and 41% report a significant increase. Industry leaders recognise both the opportunities and dangers associated with this rapid expansion. While 90% believe AI has the potential to increase bias and digital exclusion among vulnerable customers, only 23% feel confident their implementation approach carries little or no risk.
The consequences are already being felt by consumers. More than a third (35%) of vulnerable customers said they could only reach a human adviser after significant effort, such as navigating automated systems or enduring long waits, while 15% were unable to reach a human at all. Only 23% reported being able to access human support easily.
Many customers also report that AI-powered systems fail to resolve their issues effectively. More than half (52%) say automated systems rarely or never solve their problem without requiring escalation to a human agent, and only 2% say automated services consistently provide the answers they need.
The emotional impact of these experiences can be significant. Nearly six in ten (58%) customers report feeling frustrated, while 33% feel isolated and unable to access the support they need. Around one-third (32%) describe being trapped in an “AI doom loop”, repeatedly redirected through automated systems without reaching a resolution.
Importantly, customers are not opposed to AI itself. Their frustrations stem primarily from poor implementation. Many feel their circumstances are not properly understood by automated systems, forcing them to repeat information and eventually seek human assistance anyway. Looking ahead, almost half (48%) worry that increasing reliance on AI will make it harder to contact a human adviser, while 44% fear it will become more difficult to obtain support for complex financial issues.
Financial services leaders recognise many of these risks. The most commonly identified concerns include algorithmic bias and unfair decision-making (41%), increased exposure to fraud and scams (41%), and reduced access to human support (41%). However, awareness is not always translating into action.
For example, while 40% of organisations identify digital exclusion as a major risk for vulnerable customers, only 24% actively assess that risk during AI implementation. Similarly, just 29% build escalation routes to human support into their systems. More broadly, many organisations have yet to embed key customer service principles into their operations. Across areas such as understanding customer needs, defining successful outcomes, testing services against real-world scenarios, monitoring performance and adapting based on results, only around four in ten organisations say these practices are fully established.
The shortcomings are even more apparent in AI development and governance. Only 31% of firms conduct sandbox testing to identify biased or unethical outcomes before deployment. Just 27% test systems using scenarios involving vulnerable customers, while only 26% carry out formal impact assessments focused on these groups.
Despite these gaps, confidence among senior leaders remains high. More than 90% believe vulnerability receives sufficient attention at board level, and 88% see AI as a major opportunity to improve customer outcomes. In fact, 82% of organisations have already introduced AI-enabled solutions specifically designed to support vulnerable customers.
Leaders highlight several promising applications. These include digital assistants that help customers complete forms and find relevant information, personalised support journeys, intelligent triage systems that direct customers to appropriate services, conversational analytics that identify vulnerability earlier, and tools that automatically simplify communications by adjusting reading age. Such technologies can reduce friction in customer journeys, improve accessibility and free human advisers to focus on situations requiring empathy, judgement and more nuanced support.
However, many organisations remain uncertain about how to implement AI responsibly. A third (33%) of leaders say they do not know who should be accountable for AI-driven outcomes affecting vulnerable customers. Similar proportions admit uncertainty about testing AI against vulnerable customer scenarios, preventing bias, and measuring whether AI is genuinely improving outcomes.
Regulatory uncertainty is another significant challenge. Although the FCA has indicated that existing regulations, including Consumer Duty, apply to AI-enabled services, firms say there is insufficient practical guidance on how these requirements should be implemented.
As a result, 85% of organisations have delayed, or are considering delaying, AI initiatives because of regulatory uncertainty. Those delays are affecting innovation, customer service improvements and commercial opportunities. Among firms that have postponed AI projects, many cite slower product development, reduced ability to improve customer outcomes and missed business opportunities.
There is strong support across the sector for greater regulatory direction. The research found that 85% of firms support the Treasury Select Committee’s recommendation that the FCA publish practical guidance on AI by the end of the year. Many believe such guidance will be critical to ensuring AI can deliver its benefits without disadvantaging the customers who need support most.
Debra Maxwell, ArvatoConnect CEO, said: “AI has enormous potential to improve customer experiences, particularly for identifying vulnerable customers, who often struggle to access timely, personalised support.
“But we’re seeing organisations lead with the technology, rather than the outcome. AI should be there to enable better experiences, not define them. That means going back to customer service 101. Understanding where customers struggle, what good looks like for them, and then designing services – with or without AI – around those needs.
“For a sector that widely recognises the risks, the gap between knowledge and mitigation is considerable.”
