Share
The UK’s HM Revenue and Customs (HMRC) is deploying artificial intelligence to crack down on tax fraud and administrative errors through a landmark £175 million deal with British tech firm Quantexa.
The 10-year partnership aims to modernise the department’s performance by utilising AI to process vast quantities of data more efficiently than human teams alone.
The primary benefit of the new system lies in its “Decision Intelligence” capabilities. Quantexa’s technology will merge internal HMRC records with external data sources to map out hidden networks of individuals and companies masking fraudulent activity.
Beyond criminal investigations, the AI will also assist with administrative efficiency by identifying unintentional errors in tax returns and matching legitimate payments that were submitted with incorrect reference numbers.
By automating these complex data-linking tasks, the government expects to recover significant lost revenue. The investment also aims to alleviate pressure on customer service staff, potentially reversing the trend of poor response times that led to over 93,000 complaints in the 2024-2025 period.
However, the move is not without potential pitfalls. Critics of government AI often point to the “black box” problem, where automated systems make life-altering financial decisions without clear reasoning.
To mitigate this, Quantexa CEO Vishal Marria emphasised that the technology is designed to support, rather than replace, human judgment. Every automated flag will still require a human audit to ensure transparency and accountability.
Data security also remains a critical concern. To address this, the firm has pledged that tax records will never leave the HMRC environment and that staff working on the project will be strictly siloed. If successful, the deal could serve as a blueprint for “digital sovereignty,” proving that the UK can rely on domestic tech firms rather than depending on Silicon Valley giants for critical national infrastructure.
Related Posts
Discover more from Tech Digest
Subscribe to get the latest posts sent to your email.