AI and the regulator: a sober reckoning about efficiency, risk, and the limits of technology

The Financial Conduct Authority has long presented itself as a steward of both market integrity and consumer protection. Its latest annual report foregrounds a striking shift in the way the organisation thinks about its workload, its processes, and the speed at which it can respond to new threats. The adoption of artificial intelligence within the FCA is not merely a matter of cutting processing times or chasing headlines about efficiency improvements. It signals a broader strategic bet: that a data driven approach, underpinned by machine learning, can transform how supervision is conducted, how investigations are prioritised, and how swiftly firms can be onboarded once they meet the regulator’s standards. The implications are meaningful, but they are not straightforward. They invite questions about the durability of the gains, the resilience of the regulator to evolving fraud tactics, and the delicate balance between automation and human judgement in areas where errors or opacity can carry real consequences for investors and consumers.

At the heart of the report lies a seemingly simple metric: for less complex supervisory tasks, AI driven automation has reduced the time spent on cases from several hours to minutes. The numbers speak in terms of a dramatic efficiency gain. When scaled across a dataset as large as the regulator’s, such improvements could translate into a sharper allocation of resources, enabling human staff to concentrate on more intricate, high value work that demands professional scepticism and nuanced analysis. The claim is not that AI will replace human investigators; rather it is that routine, repetitive tasks can be automated, freeing up specialists to apply their expertise where it matters most. This is a familiar pattern in public administration, yet it is notable in a sector where speed, accuracy and public accountability are simultaneously demanded. The FCA is thus attempting to reconcile speed with the rigorous standards that underpin the rule of law in financial markets.

The ambitions articulated by the regulator go beyond faster case handling. The annual report describes a deliberate plan to extend AI enabled efficiencies to authorisations procedures, potentially speeding up the onboarding of firms and the approval of new products or services. The logic here is straightforward: if the regulatory gatekeeping process can be accelerated without compromising safety and compliance, more legitimate ventures will reach the market with fewer bottlenecks. In a sector prone to mis selling and mis selling detriment, an efficient authorisations regime could have a positive impact on competition and consumer choice by reducing unnecessary frictions while maintaining robust safeguards. Yet the regulator recognises that speed must not outpace scrutiny. The same document that celebrates faster processing also acknowledges the need to monitor for bias in data, to ensure transparency in decision making, and to keep faith with the public that automation is a supplement to, not a substitute for, human oversight.

The context in which these technological advances are unfolding is instructive. The FCA has been navigating a period of significant upheaval and reputational recalibration. The leadership of Nikhil Rathi, who has steered the organisation since 2020, is being tested not only by the scale of the reforms but also by the reputational bruises common to a body that has to police some of the most complex and fluid markets in the world. The annual report frames the transformation as a long term project, recognising that technology is a tool within a wider institutional evolution. The regulator accepts that while automation can improve efficiency, it cannot by itself deliver the robust enforcement and informed policymaking that are essential to maintaining public trust. In this sense the report reads less like a triumphal brochure for a new toolkit and more like a sober assessment of what has been achieved and what remains to be done.

There is a countervailing thread in the document that should not be ignored. The FCA cites the persistence of fraud as a major threat, with losses from investment fraud rising sharply. The figures are stark: losses nearly doubled from 2024 to the following year, moving from 552.6 million pounds to 1.2 billion pounds. These numbers do more than illustrate a problem in isolation; they illuminate a broader dynamic in which technological capability is both a shield and a potential catalyst for new kinds of crime. On one hand, AI and data analytics can help the regulator identify patterns, flag suspicious activity and accelerate investigations into large, complex schemes. On the other hand, criminals adapt rapidly, and the same digital tools that enable quicker detection can also empower more sophisticated fraud networks. The regulator is thus confronted with a continuous arms race, a reality that tempers any celebratory narrative about automation with a reminder that defensive technologies must be matched by proactive public education and ongoing capacity to respond to evolving threats.

The same tension is evident in the report’s discussion of market abuse and unusual trading around takeover announcements. The preponderance of suspicious trading activity preceding these announcements raises critical questions about insider dealing and information leakage. The regulator is candid in acknowledging that while AI can enhance the capacity to detect patterns, the roots of such wrongdoing lie in human behaviour and networks that stretch beyond the capabilities of any one technology. The upshot is that the FCA must maintain a two pronged approach: leverage machine driven analysis to spot anomalies, while simultaneously strengthening the institutions and integrity of markets so that information flows cannot be exploited with impunity. In other words, technology is a powerful instrument, but it cannot substitute for a robust culture of compliance, rigorous governance within firms, and a reliable set of deterrents that make misconduct less likely in the first instance.

Internal dynamics within the organisations supervised by the FCA are another area where the report offers insight. The regulator has observed an upturn in whistleblowing reports, with the total rising by 22 percent to 1,375 in the year to March. The increase in tip offs is not itself an unambiguous good or bad sign; it foregrounds the importance of internal governance within firms and the willingness of employees to raise concerns. Whistleblowing can generate valuable intelligence that leads to direct action, as the FCA reports in relation to the 523 instances of such action that followed. The interpretation of these figures is nuanced. A rise in whistleblowing could indicate greater trust in the regulatory framework and a healthier cultural willingness to report wrongdoing. Alternatively it could reflect underlying misconduct that has become more prevalent or perceived as more pervasive, potentially pointing to a need for stronger internal controls or a more responsive enforcement regime. The regulator’s challenge is to convert whistleblowing into timely, effective enforcement while continuing to improve the conditions inside organisations that encourage ethical conduct.

The report also engages with a high profile and controversial area of public policy: the redress scheme for motor finance mis selling, a multi year, multi billion pound commitment designed to compensate consumers who suffered losses. The scale of this scheme, with a headline figure of 9.1 billion pounds, has produced legal challenges from some motor finance companies. The FCA chairs a delicate balancing act here. On the one hand, a swift, fair and comprehensive redress process is essential to preserving consumer confidence and market legitimacy. On the other hand, the regulator must avoid creating incentives for opportunistic practices that could undermine the integrity of lending markets or overburden smaller firms with unmanageable liabilities. The leadership has acknowledged the obstacles in moving this project forward while maintaining a credible timetable for resolution. In this sense, the FA has to police not only the behaviour of regulated firms but also the expectations of a wider public that looks to it to deliver redress with due speed and care.

Turning to remuneration, the report provides a reminder that the public administration landscape is not solely about outcomes but also about accountability and the judgements of those who lead large organisations. The pay packages of senior officials, including the chief executive, reflect both the responsibility of stewardship and the scrutiny that accompanies such positions. The disclosed figure for Nikhil Rathi’s total compensation offers a window into the economics of leadership in a public regulatory body. It is a figure that will be interpreted through lenses of fairness, performance, and the broader debate about public sector remuneration. The emphasis of the document, however, remains on strategic and operational transformation rather than on individual metrics.

The regulatory agenda outlined in the annual report is not a simple linear project. It recognises that the adoption of AI is not a panacea that will automatically resolve all issues of market misconduct or consumer harm. The technology’s effectiveness is contingent on the quality, depth and diversity of data, and the safeguards that govern how decisions are made. The so called black box problem is not merely a theoretical concern; it sits at the heart of questions about explainability, accountability and the potential for bias to seep into automated decisions. The FCA acknowledges these concerns, signalling an awareness that there are limits to what automation can deliver and a clear commitment to address those limits through governance, human oversight, and transparent reporting.

The broader implications for the functioning of financial markets are clear, even if the path ahead remains uncertain. A regulator that can process routine tasks with speed and accuracy has more capacity to monitor, intervene and respond when mispricing or mis selling threatens to destabilise confidence. The same regulator that is embracing AI must remain vigilant against the risks that accompany new technologies, including the inadvertent amplification of biases in data sets, the opacity of automated decision making, and the challenges of maintaining lines of accountability when decisions are increasingly mediated by algorithms. The regulator must also navigate public expectations that it will deliver not only faster outcomes but also better outcomes, ensuring that the benefits of technological progress translate into tangible improvements for ordinary people who rely on the integrity of financial markets.

If there is a broader takeaway from the FCA’s current trajectory, it is that the relationship between technology and regulation is best understood as a work in progress rather than a finished project. The regulator is testing and refining new tools in real time, learning from both successes and missteps. This iterative approach echoes the nature of modern regulation itself: adaptive, evidence driven and capable of adjusting course as markets evolve and new threats emerge. The key, as the report implies, is not simply to deploy more technology but to cultivate the governance and ethical frameworks that ensure technology serves the public interest. When automation is implemented with robust oversight, where human judgement remains central to interpretation and decision making, AI can become a genuine force multiplier for a regulator that must police some of the most complex markets in the world.

At the same time, the report’s sober tone underscores the limits of what any regulator can achieve alone. Fraudsters and market manipulators constantly adapt to new technologies, and public policy must keep pace with those adjustments. The FCA’s effort to balance efficiency with accountability, speed with caution, innovation with safety, is a demanding enterprise. The ultimate measure of success will not be the height of the efficiency gains reported in the most optimistic passages, but the regulator’s ability to translate those gains into fewer victims of fraud, swifter redress for harmed consumers, and clearer protections for investors in a rapidly changing financial landscape.

In the end, the FCA’s current approach is not a radical departure so much as a default setting that recognises the realities of modern market supervision. Technology can transform processes, but it cannot replace the need for robust governance, human judgement and a continuous, honest assessment of risk. The regulator’s ambition to harness AI while preserving the core principles of accountability and fairness is a timely reminder that the most valuable asset in any system designed to police powerful interests is not merely clever software, but an intelligent, principled and brave public institution prepared to adapt without compromising its obligations to the public it serves.