Smile for the taxman – HMRC’s AI is watching

In a move that has ignited significant debate on data privacy and the role of artificial intelligence in public service, HM Revenue & Customs (HMRC) has confirmed it is using AI to monitor taxpayers’ social media posts.

While the tax authority insists this practice is currently limited to criminal investigations and is supported by “robust safeguards,” the revelation has raised a storm of questions about the future scope of AI in tax enforcement and the potential for a “Horizon Post Office-type scandal.”

The new reporting, sourced from multiple UK outlets, confirms that HMRC’s AI tools are being used to analyze social media posts, along with financial records and spending habits, to identify discrepancies that could indicate tax evasion. The system flags “red flag” purchases like expensive holidays or luxury items if they appear inconsistent with a person’s declared income. This marks the first time HMRC has publicly admitted to this specific application of AI.

This AI-driven surveillance exists alongside a separate, long-established system called Connect, which uses billions of data points to conduct routine tax investigations. However, the new focus on social media and the planned expansion of AI into “everyday” tax processes signals a significant shift in HMRC’s strategy. With a target of closing a £7bn portion of the UK’s £47bn “tax gap,” the reliance on AI is seen as a way to increase efficiency and reduce the need for human investigators.

Bob’s Take: The Privacy Paradox and the “Trust” Problem

“This development presents a classic privacy paradox for financial services and fintech. On one hand, the use of AI to combat tax evasion is, in principle, an efficient way to ensure fairness and generate revenue for public services. On the other hand, the move to monitor personal social media posts—a space traditionally considered private—erodes the public’s trust in the government’s handling of personal data.

For financial institutions, this raises questions about the data they share with government bodies. While banks are already obligated to share data with HMRC, the expansion into social media monitoring sets a new precedent for what constitutes a “digital footprint” relevant to financial oversight. This is especially pertinent to our audience, where data protection regulations like GDPR and various state-level privacy laws are fiercely debated. The move by the Labour government to amend HMRC’s privacy policy, shifting from a guarantee of “human judgement” to an assurance of “human involvement,” is a subtle but crucial change. This could be the thin end of the wedge, paving the way for more automated, less scrutinized decisions.”

Is AI the New Informant?

The most sensitive and rarely discussed angle in this story is the potential for AI to act as a state-sanctioned digital informant, blurring the lines between crime-fighting and surveillance. While HMRC insists the social media monitoring is for criminal cases only, the advanced nature of the technology means it can easily be scaled.

The “Horizon Post Office-type scandal” comparison, raised by senior MPs, is no exaggeration. The core of that scandal was the blind trust placed in an automated system, leading to the wrongful prosecution of innocent people. A flawed AI model could misinterpret a social media post, flag an innocent person for a “red flag” purchase that was a gift or a loan, and trigger an automated investigation. The danger is that the system’s “explainability” and “human involvement” safeguards are not robust enough, or that human agents are pressured to trust the AI’s findings without proper scrutiny.

The fear is not just about tax evasion, but about the chilling effect on freedom of expression.

Will people self-censor their social media posts to avoid the possibility of a tax audit?

Will they hesitate to share good news about a new car or a big holiday out of fear of being flagged by an algorithm?

This moves the discussion from a technical cybersecurity issue to a fundamental question of civil liberties in a digital age. It’s a conversation the UK needs to have now, before AI becomes so deeply embedded in government processes that rolling it back becomes impossible.