The technology flags up individuals who appear to be living beyond their declared means, for example by posting about expensive holidays or large purchases.

An HMRC spokesman insisted the tools are used with “robust safeguards in place” and only for monitoring social media in criminal cases, a practice that has reportedly been in place for several years and is within the law.

This use of AI exists alongside ‘Connect’, a powerful IT system that has been examining billions of financial data points for over a decade to spot signs of tax avoidance. The increasing reliance on technology is seen as key to HMRC’s efforts to close the UK’s £47 billion “tax gap”, The Telegraph reports.

However, the admission has sparked alarm among senior MPs, who fear that wider use of AI without human oversight could lead to errors and unfair penalties. Sir John Hayes, a former security minister, warned against automated processes, stating, “the idea that a machine must always be right is what led to the Post Office scandal”. Bob Blackman MP described the potential for legal action based on AI findings as “draconian”.

Concerns have been amplified by HMRC’s future plans. A recent report revealed intentions to use AI to identify suspected tax evaders and send “automated nudges” asking them to pay what they owe. This move towards automation follows a change to HMRC’s privacy policy in May. A previous statement guaranteeing that AI “does not replace human judgement” was removed and replaced with a commitment to ensure “human involvement” where AI could impact customer outcomes.

Labour has indicated it hopes to recoup £7bn of the tax gap, and improvements to AI software could be central to achieving this goal.