Closing the Tax Gap with Transformational AI to Strengthen Fraud Prevention and Protect Public Revenue

Guest article, by Amanda Reynolds, Managing Director, Accenture and Heather Adams, Global Head Risk & Compliance Consulting, Accenture.

Fraud, including tax evasion and tax avoidance, is a significant contributor to the UK’s tax gap. Billions that could fund essential public services is not collected each year. These losses reduce funding available for essential public services and create unfair advantages for those who break the rules over those who comply. Preventing fraud is therefore not only an operational priority for UK Government; it is a foundational requirement for maintaining the integrity, fairness, and resilience of the tax system.

Since 2005-2006, the tax gap has been falling, as compliance and anti-fraud work has taken effect.¹ However, sustaining this is challenging in the UK where fraud (across all sectors is on the rise.² To address this, government departments are turning to artificial intelligence to enhance their capability to detect, deter, and ultimately prevent fraudulent activity at scale.

Fraud prevention in a rapidly evolving environment

Tax fraud today is more agile and sophisticated than ever. Fraudsters exploit new technologies, complex financial structures and cross-border mechanisms to obscure their activities. Traditional investigative methods, while still essential, cannot keep pace with the speed and complexity of these schemes. ³

AI is as critical a tool for those in fraud prevention as it is for the fraudsters. By scanning vast quantities of data at speed, AI systems offer UK government the ability to detect subtle behaviours and emerging patterns that indicate a heightened risk of fraudulent activity - often long before a human investigator would spot them.

This shift from reactive detection to proactive prevention marks one of the most transformative developments in modern tax administration.

AI as an early-warning system

AI can act as an advanced early-warning system. By analysing historical cases, income patterns, behavioural anomalies and network connections, AI can identify irregularities with exceptional accuracy.

For example, AI can flag inconsistencies in reported income versus lifestyle indicators, detect unusual transactional flows, or identify clusters of suspicious activity across multiple taxpayers. These insights enable early intervention, before fraud escalates or repeated non-compliance accumulates into significant revenue loss. Early identification is central to effective prevention, and AI strengthens the ability to act decisively and at scale.

AI also helps investigators focus on the right cases. By learning from experience, machine learning models get better over time, reducing false alarms and freeing up time for genuine risks to be reviewed. This precision means faster investigations, stronger outcomes and earlier disruption of fraudulent schemes, all while reducing unnecessary pressure on honest taxpayers.

Enhancing investigation precision

Preventing fraud also requires precise targeting. AI reduces the number of false positives that can consume investigative time and burden legitimate taxpayers. By continuously learning from new cases and feedback, machine learning models refine their predictions, ensuring that attention is focused on the highest-risk individuals and businesses.

This improved accuracy not only accelerates investigations but also increases the likelihood of successful interventions, penalties and recovery actions. More importantly, it allows UK government to disrupt fraudulent schemes earlier in their lifecycle which significantly reduced exposure to loss.

Strengthening operational capacity to combat fraud

Fraud prevention is not just about finding anomalies; it is about building the organisational capacity to address them. AI helps by automating labour-intensive tasks such as data ingestion, document verification and cross-checking records.

This operational uplift frees experienced investigators to focus on high-complexity cases where specialist judgement is crucial. It also enables more cases to be pursued simultaneously, reducing the opportunity for fraudsters to exploit processing delays or administrative backlogs.

Predictive and preventative capabilities

The next big leap will be to use AI in a way that is more predictive, running multiple what if scenarios and predicting their impact. AI systems can forecast emerging risk areas based on trends, seasonal patterns and behavioural shifts. This allows UK government to implement proactive strategies, allocate resources more effectively and engage taxpayers before non-compliance occurs.

Digital twins could also predict how taxpayer behaviours might change in response to changing beliefs or circumstances. This intelligence can enable targeted communication, early intervention, and tighter controls, significantly reducing the window of opportunity for fraud to take hold.

Leveraging unstructured data for deeper insight

Generative AI can read and interpret unstructured data like emails, contracts or reports to uncover patterns and connections that would otherwise go unnoticed. AI can combine insights from both internal and external data (structured and unstructured) to identify risk indicators that would be difficult or impossible to detect manually.

These capabilities broaden visibility into potential fraud schemes, helping uncover hidden relationships and previously undetected patterns of behaviour.

Collaboration and continuous evolution

Preventing fraud requires constant adaptation. As fraudsters evolve their methods, we must evolve our tools. Collaboration between UK government and partners enables AI solutions to be updated, refined and evaluated against emerging threats. These partnerships combine domain expertise, technological innovation and operational insight to build solutions that keep pace with the highly dynamic nature of tax fraud.

A more secure and fair tax system

AI is not just about improving operational efficiency; it’s fundamentally strengthening the prevention of tax fraud. By identifying risks earlier, analysing data more comprehensively and supporting more targeted interventions, AI helps protect public revenue, reinforce fairness and maintain trust in the tax system.

As government continues to invest in advanced technologies and collaborative innovation, the UK is better positioned to close the tax gap, safeguard the tax base and ensure that taxpayers who play by the rules are not disadvantaged by those who do not.

References:

1. HMRC performance data 2025 to 2026: October - GOV.UK
2. NSA 2025 - Fraud - National Crime Agency
3. INTERPOL Financial Fraud assessment: A global threat boosted by technology

Heather Adams will be hosting a Seminar at Counter Fraud 2026, entitled: "Using AI to streamline fraud detection and prevention and improve customer service". Book your free public sector ticket to join her, live, on 26th February at QEII Centre, Westminster >> https://www.counterfraudconference.co.uk/agenda