Three-quarters of vulnerable customers give up trying to reach their bank
Financially vulnerable people are hitting walls when they try to get help from their bank, insurer, or credit provider. A new survey of 1,000 senior decision-makers across UK financial services firms found that 74% of vulnerable customers have felt like giving up while seeking support, and 26% abandoned their attempt entirely.
The culprit: poorly designed AI for Customer Support systems that are making access worse, not better.
Banks know the risks but are deploying AI anyway
More than three-quarters of finance leaders (77%) admit their AI strategies risk harming vulnerable customers. Yet almost nine in ten (88%) have increased their use of AI in customer-facing operations over the past 12 months, with 41% reporting a significant increase.
The pressure to adopt AI is outpacing the ability to implement it safely. Only 23% of leaders feel confident their approach poses little or no risk. Just 6% believe their approach poses no risk at all.
Vulnerable customers trapped in automated loops
The impact is already visible. When vulnerable customers contact their provider for help, 35% can only reach a human after significant effort navigating automated systems and long waits. Another 15% cannot reach a human at all.
Automated systems rarely resolve issues without escalation to a human, according to 52% of vulnerable customers. Only 2% say automated systems consistently provide the answers they need.
A third of vulnerable customers (32%) describe feeling trapped in an "AI doom loop"-repeatedly redirected through automated systems without resolution. Many report feeling frustrated (58%) or isolated and unable to get support (33%).
The gap between knowing risks and preventing them
Finance leaders identify the top risks of AI as algorithmic bias (41%), increased fraud exposure (41%), and reduced access to human support (41%). Yet these concerns rarely translate into action during system design.
Only 24% of organisations assess digital exclusion risk during AI implementation, even though 40% recognise it as a major threat to vulnerable people. Just 29% build in escalation routes to human agents.
The fundamentals of good customer service design are inconsistently applied. Around four in ten organisations say they fully embed practices like understanding customer needs, defining successful outcomes, testing against real scenarios, monitoring performance, and adapting based on results.
Testing for problems before deployment is rare. Only 31% sandbox-test AI systems for bias or unethical outcomes. Just 27% test using vulnerable customer scenarios, and 26% conduct formal impact assessments on vulnerable populations.
Accountability unclear
A third of leaders (33%) do not know who should be accountable for AI-driven outcomes affecting vulnerable customers. Similarly, 32% do not know how to test AI against vulnerable customer scenarios, 31% are unclear on preventing bias, and 31% cannot measure whether AI is improving outcomes for these customers.
Regulatory uncertainty is slowing rollout
The Financial Conduct Authority has indicated that existing frameworks, including Consumer Duty, should apply to AI, but has not provided detailed guidance on implementation.
As a result, 85% of firms have delayed or are considering delaying AI implementation due to lack of clarity. Among those who have delayed, consequences include slower product development (33%), reduced ability to improve customer outcomes (31%), and missed commercial opportunities (30%).
In the absence of clear UK guidance, firms are adopting a patchwork of frameworks: internal governance models (47%), ISO 42001 (42%), the US NIST AI Risk Management Framework (37%), the EU AI Act (30%), and other international standards (24%).
There is overwhelming support for clearer direction, with 85% of firms backing the Treasury Select Committee's recommendation that the FCA publish practical AI guidance by year-end.
Where AI could actually help
Despite the shortcomings, 88% of finance leaders believe AI can have a positive impact on vulnerable customers. Eighty-two percent have already implemented AI-enabled solutions designed to support them.
Leaders see promise in ChatGPT-style digital agents that assist with form-filling and signposting (36%), personalised support journeys with intelligent triage (35%), conversational analytics to identify vulnerability earlier (34%), and tools that adjust reading age of communications (30%).
These tools can remove friction by guiding customers through forms and helping agents provide the right information first time. They can also free up human agents to focus on cases requiring more empathy or nuance.
The gap between potential and practice remains wide. Firms that focus on strong customer service fundamentals-clear escalation routes to humans, robust testing, and designing around customer outcomes-will be better positioned to implement AI responsibly under existing Consumer Duty expectations.
Your membership also unlocks:
