Three-quarters of financial services leaders believe their own AI strategies could harm vulnerable customers, as new research reveals growing concerns over access to human support and widening digital exclusion.
Financial services firms risk making it harder for vulnerable customers to access support as AI adoption accelerates across the sector, according to new research from ArvatoConnect.
The study, which surveyed 1,000 senior decision-makers across U.K. banks, insurers, fintechs, building societies and credit providers, found widespread concern among leaders that AI-driven customer service strategies are advancing faster than the safeguards needed to protect vulnerable customers.
Around 77 percent of financial services leaders said their organisation’s AI strategy could have a negative impact on vulnerable customers, with 28 percent describing the risk as high.
The findings come as AI deployment continues to gather pace. Around 88 percent of organizations have increased their use of AI in customer-facing operations over the past year, including 41 percent that reported a significant increase.
At the same time, financially vulnerable consumers continue to face significant barriers when seeking help from providers.
According to the research, 74 percent of vulnerable customers have felt like giving up while trying to get support from a bank, insurer or financial provider, and 26 percent have abandoned attempts to seek help altogether.
Human Support Remains Difficult To Access
The research points to ongoing challenges in balancing automation with access to human assistance.
Around 35 percent of vulnerable customers said they could only reach a human adviser after considerable effort navigating automated systems and waiting queues. A further 15 percent reported being unable to reach a human at all. Just 23 percent said they could access a human adviser easily.
Automated systems also appear to be falling short in resolving customer issues. Around 52 percent of customers said that AI or automated systems rarely or never solved their problem without escalation to a human agent, while only 2 percent reported consistently receiving the answers they needed through automated channels.
The experience is taking an emotional toll, with 58 percent of vulnerable customers reporting feeling frustrated during support interactions, while one-third said that they felt isolated and unable to access the help they needed.
Around 32 percent described becoming trapped in an “AI doom loop,” repeatedly being redirected through automated systems without reaching a resolution.
Awareness of Risks Not Translating Into Action
While financial services leaders appear highly aware of AI-related risks, the research indicates that their organizations are lagging behind in implementing measures to limit the impact on vulnerable customers.
The vast majority, 90 percent, believe that AI could worsen bias and digital exclusion among vulnerable customers. Yet only 23 percent expressed confidence that their organization’s implementation approach presents little or no risk.
Leaders identified algorithmic bias, increased exposure to fraud and scams and reduced access to human support as the three most significant risks associated with AI adoption, each cited by 41 percent of respondents.
However, the responses suggest that these concerns are not consistently reflected in implementation practices. While 40 percent of organizations recognise digital exclusion as a major risk for vulnerable customers, only 24 percent assess that risk during AI deployment projects. And just 29 percent build escalation routes to human support into AI-enabled journeys.
The same pattern appears across customer experience design more broadly. Only around four in ten organisations said key service design disciplines, including understanding customer needs, defining successful outcomes, testing customer journeys and continuously improving services, are fully embedded within their operations.
Debra Maxwell, CEO of ArvatoConnect, said organisations need to focus on customer outcomes rather than technology deployment.
“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.”
Governance Gaps Emerge in AI Deployments
The research also reveals significant uncertainty around responsible AI governance.
Only 31 percent of organizations said they sandbox-test AI systems for biased or unethical outcomes before deployment. Just 27 percent test systems using vulnerable customer scenarios, while 26 percent conduct formal impact assessments focused on vulnerable groups.
Knowledge gaps remain widespread among senior leaders. Around 33 percent said they do not know who should be accountable for AI-driven outcomes involving vulnerable customers.
Meanwhile, 32 percent said they are unsure how to test AI tools using vulnerable customer scenarios before deployment, while 31 percent admitted uncertainty about preventing bias or measuring whether AI is improving outcomes.
Despite these challenges, confidence remains high. Around 92% percent believe vulnerability receives sufficient attention at board level.
Opportunity To Improve Support Remains Significant
The findings also indicate strong optimism about the potential for AI to improve accessibility and support when it is implemented effectively.
Around 88 percent believe AI can deliver positive outcomes for vulnerable customers and 82 percent said their organization has already introduced AI-enabled tools designed specifically to support them.
Among the most promising use cases identified were digital assistants that can help customers complete forms and access information, personalized support journeys, intelligent triage systems, conversational analytics that identify vulnerability earlier and tools that automatically adjust the reading age of communications.
The findings suggest the challenge for customer experience leaders is deploying AI in ways that improve outcomes while preserving access to human support where needed.
