UK Tax Authority Embraces AI to Sharpen Fraud Detection and Error Correction
By Tom Gray
•
Published: May 18, 2026
HM Revenue and Customs (HMRC), the UK's tax authority, is set to revolutionize its approach to tackling tax fraud and errors by implementing advanced artificial intelligence (AI) technology. The department has awarded a substantial £175 million contract to Quantexa, a British financial data platform specialist, to significantly enhance its capabilities in identifying discrepancies and fraudulent activities within tax returns. This ten-year partnership marks a pivotal shift towards modernizing HMRC's data infrastructure and leveraging sovereign AI at a national scale.
The agreement, announced on May 14, 2026, is poised to be one of the largest Decision Intelligence projects ever undertaken within the UK government. Quantexa's state-of-the-art platform will work to unify and connect fragmented data sets that currently reside in various legacy systems across HMRC. This unification will create a comprehensive, trusted view of information, providing sharper insights into potential tax risks and bolstering compliance controls.
A primary function of the deployed AI will be to identify sophisticated tax fraud and evasion schemes, including hidden networks of companies and individuals engaged in illicit activities. The technology, which utilizes graph analytics and machine learning, is designed to trace complex relationships between entities across multiple layers of corporate ownership – a task that would be virtually impossible for human auditors using traditional methods. Beyond fraud, the system will also be instrumental in spotting unintentional errors in tax returns and even tracking legitimate payments made with incorrect reference numbers.
Quantexa, a London-based company founded in 2016, specializes in developing AI applications and tools for data analytics and decision-making. Their Decision Intelligence Platform has a proven track record in financial crime and anti-money laundering within the banking sector, helping institutions like HSBC detect suspicious transactions. This expertise will now be applied to the complex landscape of tax administration, helping HMRC to target 'high-probability' fraudulent activities.
The implementation of this AI system is expected to assist HMRC customer service operations by helping staff resolve queries more swiftly, potentially improving public satisfaction which has seen a rise in complaints in recent years. Crucially, Quantexa has emphasized that its AI systems are intended to support human decision-making, not replace it, with all AI-generated findings subject to human review before any official action is taken. Quantexa's CEO, Vishal Marria, highlighted the importance of transparency and audibility, stating that in government environments, AI decisions must be explainable, particularly when affecting citizens directly. The company has also assured that HMRC's data will remain securely within its own environment.
This strategic investment forms part of a broader government initiative to close the UK's "tax gap" – the difference between the tax owed and the amount actually collected, which reached £46.8 billion in 2023-24. By harnessing advanced AI, HMRC aims to recover billions lost to evasion and sophisticated fraud, contributing to the government's commitment to retrieve an additional £10 billion annually by 2030. The deal also signifies a deliberate move towards strengthening British AI sovereignty within critical national functions.
The agreement, announced on May 14, 2026, is poised to be one of the largest Decision Intelligence projects ever undertaken within the UK government. Quantexa's state-of-the-art platform will work to unify and connect fragmented data sets that currently reside in various legacy systems across HMRC. This unification will create a comprehensive, trusted view of information, providing sharper insights into potential tax risks and bolstering compliance controls.
A primary function of the deployed AI will be to identify sophisticated tax fraud and evasion schemes, including hidden networks of companies and individuals engaged in illicit activities. The technology, which utilizes graph analytics and machine learning, is designed to trace complex relationships between entities across multiple layers of corporate ownership – a task that would be virtually impossible for human auditors using traditional methods. Beyond fraud, the system will also be instrumental in spotting unintentional errors in tax returns and even tracking legitimate payments made with incorrect reference numbers.
Quantexa, a London-based company founded in 2016, specializes in developing AI applications and tools for data analytics and decision-making. Their Decision Intelligence Platform has a proven track record in financial crime and anti-money laundering within the banking sector, helping institutions like HSBC detect suspicious transactions. This expertise will now be applied to the complex landscape of tax administration, helping HMRC to target 'high-probability' fraudulent activities.
The implementation of this AI system is expected to assist HMRC customer service operations by helping staff resolve queries more swiftly, potentially improving public satisfaction which has seen a rise in complaints in recent years. Crucially, Quantexa has emphasized that its AI systems are intended to support human decision-making, not replace it, with all AI-generated findings subject to human review before any official action is taken. Quantexa's CEO, Vishal Marria, highlighted the importance of transparency and audibility, stating that in government environments, AI decisions must be explainable, particularly when affecting citizens directly. The company has also assured that HMRC's data will remain securely within its own environment.
This strategic investment forms part of a broader government initiative to close the UK's "tax gap" – the difference between the tax owed and the amount actually collected, which reached £46.8 billion in 2023-24. By harnessing advanced AI, HMRC aims to recover billions lost to evasion and sophisticated fraud, contributing to the government's commitment to retrieve an additional £10 billion annually by 2030. The deal also signifies a deliberate move towards strengthening British AI sovereignty within critical national functions.