The Mills Review, named after FCA Executive Director Sheldon Mills, lays out some predictions for how AI will irrevocably alter the finance sector, locally and globally, in the run-up to the end of the decade.

The FCA claimed the “landmark” review, published in July 2026, is “the first work of its kind initiated by a regulator globally”. Although the report’s emphasis is on retail finance, it points to profound changes that are likely to impact businesses more widely.

The report points to four overarching "systemic shifts” that are likely to occur amid AI’s increasing role: 

• firms in the sector will be transformed;
• the consumer finance journey will be “agent-led”, where AI initiates and monitors transactions and investments on behalf of human ‘owners’;
• market power and competition will be reshaped; and
• technology-driven threats and defences will proliferate.

“The central shift is from human-led, episodic financial activity towards services that are AI-enabled, continuous and delegated. AI will operate inside firms, through consumer interfaces, across markets and within regulators. It will affect how products are designed, distributed, monitored and governed,” Mills concluded in the review’s executive summary. 

With those shifts providing context, here are five of the most important, direct ways in which AI is predicted to upend UK accounting and finance by the end of this decade.

1. From automation to autonomy

Between now and 2030, advanced agentic AI tools will move beyond basic robotic process automation (RPA) routines and graduate to environments where the AI operates continuously and autonomously. As this occurs, the humans working with those tools move from being the operator of those tools to gradually becoming an observer, monitoring the work that AI does. As this takes place, it in turn reshapes the regulatory risk around AI, creating the possibility of “system-wide” harms.

The humans within the firm will begin to move along what the Mills Review calls an “autonomy spectrum”. This is a  scale of five roles on which they are elevated from running tasks, to setting the parameters, granting the permissions and overseeing outcomes produced by AI agents that perform the tasks instead. The five roles include: 

• the ‘operator’, where the human uses AI as a tool;
• the ‘collaborator’, where the human and AI work together on tasks; 
• the ‘consultant’, for whom AI would weigh options and recommendations, but the human makes final decisions;
• the ‘approver’, where AI prepares the actions for the human to authorise; and
• the ‘observer’, where AI acts within set limits while the human monitors outcomes.

2. ‘Always on’ real-time monitoring

AI agents will potentially be able to track data routes continuously, identifying anomalies and potential fraudulent transactions in real time. Some financial services firms are already piloting AI in customer support and fraud detection and are shifting towards more real-time monitoring of data and behaviours.

The Mills Review outlines three governance areas that AI could support:

• executing customer operations, product delivery and servicing;
• risk, compliance, monitoring and policy; and
• internal audit, assurance and testing.

As AI increases cyber risks, AI could also be used to continuously monitor and detect attempts to breach servers and data bases, and detect system vulnerabilities.

3. Customisation of client services

Generative AI might be used to help deliver automated, data-driven financial advice. For example, AI might augment business’ localised financial data with broader macroeconomic trends with which to offer strategic investment advice.

4. Less process-oriented tasks, more governance workloads for humans

These evolving roles mean that governance becomes a key operating constraint, rather than a compliance add-on. Professionals will end up testing for system transparency, checking and testing algorithms, proving a level of transparency and understanding to clients on what can be rather opaque issues, as well as mapping clear corporate AI liabilities in line with the FCA's strict framework.

5. Technology arms race: defensive financial crime mitigation

There will be a proliferation of new AI-powered tools on both sides of the legal and criminal divide. On the one hand, AI will render corporate fraud, deepfakes and identity theft faster and cheaper. On the other, forensic accounting will respond and gear up accordingly. Teams will deploy sophisticated, AI-driven defensive systems to identify and intercept – if not neutralise – near-perfect synthetic invoice scams or ledger attacks before accounts or data are compromised.

“The Mills Review provides a valuable roadmap for understanding how AI will transform financial services, and importantly recognises that innovation must be matched by strong governance,” says Polly Tsang, Senior Financial Services Regulation Manager, ICAEW. “Our AI in Financial Services roundtable with industry and regulators two years ago highlighted many of these emerging challenges, from hyper-personalisation and consumer outcomes to accountability where firms rely on third-party technology providers.

“As AI agents become increasingly autonomous, maintaining clear human oversight and robust governance frameworks will be critical to ensuring trust in the next generation of financial services.”