The Bank of England and the Financial Conduct Authority have jointly published their latest assessment of artificial intelligence adoption across financial services — and, depending on how you read it, the report is either a triumph of technological progress or a quietly damning portrait of an industry that keeps reaching for new tools to protect old habits. The numbers, by any measure, look striking. The reality they describe, however, raises an uncomfortable question that senior executives and policymakers alike have been reluctant to answer directly: are banks genuinely transforming themselves, or are they simply making yesterday's banking a little cheaper to run?
It is worth pausing on the framing before dismissing that question as rhetorical provocation. AI adoption across the UK financial sector has accelerated at what the report characterises as extraordinary speed. The deployment of machine learning models across credit scoring, fraud detection, customer service automation, and regulatory compliance has moved from pilot projects to enterprise-scale infrastructure in a compressed timeframe that would have seemed implausible even five years ago. The investment figures, the adoption curves, and the productivity metrics cited in the report are, by the Bank of England's own description, impressive. That much is not in dispute.
What is in dispute — and what the report, perhaps inadvertently, surfaces — is the nature and direction of that transformation. There is a meaningful difference between using artificial intelligence to execute existing banking processes faster and at lower cost, and using it to fundamentally reimagine what a bank does, who it serves, and how value is created and distributed. The former is efficiency. The latter is transformation. The concern embedded in the gap between the report's headline numbers and its underlying reality is that the industry is delivering the former while marketing it as the latter.
This is not a new tension. Banking has a long institutional history of adopting powerful new technologies in ways that reinforce rather than disrupt existing structures. The arrival of mainframe computing in the 1960s and 1970s automated the back office but left the branch model and its attendant economics largely intact. The internet era of the late 1990s — the era evoked pointedly in the question "why are we banking like it's 1999?" — promised to dissolve the geographic constraints of traditional banking. What it largely produced instead was online access to the same products, the same risk models, and the same customer relationships that existed before. Convenience improved. The underlying architecture did not.
The risk today is structurally similar. A bank that deploys a large language model to handle customer enquiries more cheaply than a call centre agent has unquestionably improved its operational efficiency. But it has not changed its product set, its credit philosophy, its approach to financial inclusion, or its relationship with the customer in any meaningful structural sense. The AI is doing the old job better. The old job remains the old job. And when regulators measure adoption by counting deployments rather than examining outcomes for the end customer or the broader financial system, the impressive numbers can obscure a more modest underlying reality.
This is precisely where the Bank of England and FCA report invites closer scrutiny. Joint publications of this nature serve multiple purposes simultaneously — they inform, they signal supervisory priorities, and they set the terms of a public conversation. The fact that both institutions have chosen to frame the discussion around AI adoption at scale is significant. It suggests a regulatory posture that is, at least for now, more concerned with managing the risks of rapid AI deployment than with interrogating whether that deployment is producing structural change. That is a legitimate and important focus. Systemic risk, model risk, and the concentration of AI infrastructure among a small number of technology vendors are genuine supervisory concerns that deserve serious attention.
But the harder question — the one that the private sector's own marketing tends to obscure and that regulators are perhaps best positioned to ask clearly — is whether the financial services industry is using its AI investment to serve customers and the economy more effectively, or whether it is primarily using it to protect and entrench existing market positions. Efficiency gains that accrue primarily to shareholders, while customers continue to navigate the same product structures, the same exclusions, and the same friction points that defined banking in 1999, represent a missed opportunity of considerable magnitude.
What This Means
The Bank of England and FCA's joint AI report arrives at a moment when the industry is spending heavily, moving quickly, and generating impressive statistics. None of that should be dismissed. But the measure of AI's success in financial services cannot be adoption speed alone. It must ultimately be judged by whether the technology is being used to build a banking system that is more resilient, more inclusive, more transparent, and more responsive to the real financial needs of households and businesses — or whether it is being used, as so many technologies before it, to make the existing system marginally more profitable while leaving its foundations unchanged. The numbers are impressive. The question of what they actually represent remains, as yet, unanswered.
Written by the editorial team — independent journalism powered by Codego Press.