HMRC’s Slow March Toward AI: A Missed Opportunity for Transformation

The recent announcement that HM Revenue and Customs (HMRC) has established a “landing zone to safely exploit” generative artificial intelligence (AI) in its 2024/25 annual report sounds promising on the surface. It signals a step toward embracing cutting-edge technology to streamline operations and enhance public services. However, beneath the optimistic rhetoric lies a troubling reality: HMRC’s adoption of AI has been frustratingly slow, lagging behind both the private sector and the transformative potential of this technology. While the establishment of a dedicated AI framework is a welcome move, it’s hard to shake the sense that HMRC is playing catch-up in a race where others have already taken the lead.

A Decade of Hesitation

HMRC’s journey with AI is not new. As early as 2017, the organisation was exploring automation and machine learning, with initiatives like the Automation Delivery Centre aiming to automate 10 million processes by 2018. The Connect system, in use for over a decade, has been a cornerstone in combating tax evasion by aggregating data from diverse sources like the Land Registry and online marketplaces. Yet, despite these early forays, HMRC’s progress in leveraging more advanced AI, particularly generative AI, has been glacial. The private sector—think tech giants, financial institutions, and even mid-sized startups—has been harnessing AI for years to optimise processes, personalise customer experiences, and drive innovation. Meanwhile, HMRC’s efforts have felt like cautious tiptoeing rather than bold strides.

The Public Accounts Committee’s (PAC) recent warning underscores this lag, stating that HMRC is “not well-placed to take advantage of the opportunities offered by technology, such as the development of artificial intelligence.” This critique stings because it highlights a systemic issue: a lack of urgency and strategic vision. While HMRC touts its “landing zone” for generative AI, the reality is that this framework feels like a belated response to a technological revolution that’s already reshaping industries worldwide.

The Promise of AI for HMRC

The potential for AI to transform HMRC’s operations is immense. Generative AI, capable of producing text, summarising data, and even generating code, could revolutionise how HMRC interacts with taxpayers. Current use cases, such as summarising customer helpline calls or powering chatbots for online advice, are promising but limited in scope. Imagine a world where AI-driven analytics predict tax evasion with pinpoint accuracy, where chatbots provide real-time, personalised guidance to taxpayers, or where AI streamlines recruitment and compliance processes to save time and resources. HMRC’s own report highlights tools like “Skill Scribe,” which simplifies recruitment, and a regional insights tool for labour market analysis—proof that the technology can deliver when applied.

Yet, these initiatives are still in their infancy. The department’s participation in a cross-government AI chatbot pilot and its use of AI for enhanced compliance targeting and fraud detection are steps in the right direction, but they feel like pilot projects rather than a comprehensive strategy. The fact that only 7,225 staff members completed AI-focused training in 2024/25, while commendable, is a drop in the bucket for an organisation of HMRC’s size. Scaling up AI adoption requires not just training but a cultural shift—one that embraces risk-taking and innovation over bureaucratic caution.

The Risks of Moving Too Slowly

HMRC’s slow adoption of AI isn’t just a missed opportunity; it’s a risk to its mission. Tax evasion and fraud are evolving, with criminals leveraging AI to create sophisticated scams, from phishing to voice cloning. A 2025 government report on the safety and security risks of generative AI warns that criminals are adopting the technology at the same pace as the general population, amplifying threats like fraud and data harvesting. If HMRC doesn’t accelerate its AI capabilities, it risks being outmanoeuvred by those it seeks to regulate.

Moreover, the public expects more from government services in the digital age. Taxpayers, accustomed to seamless experiences from private-sector platforms, are frustrated by clunky processes and long wait times. AI could alleviate these pain points—scheduling appointments, personalising taxpayer support, or automating repetitive tasks to free up staff for complex cases. But every year HMRC delays, it erodes public trust and misses chances to improve service delivery.

Ethical Concerns and the Need for Speed

To its credit, HMRC is taking steps to address ethical concerns, with an AI Ethics Working Group overseeing mandatory processes and ensuring transparency. This cautious approach is necessary—AI misuse could lead to biased outcomes or privacy breaches, as highlighted by the Information Commissioner’s Office and the UK’s AI governance frameworks. However, caution shouldn’t mean paralysis. The private sector has shown it’s possible to balance innovation with ethical safeguards, and HMRC must learn to do the same.

The “landing zone” concept is a step toward safe AI adoption, but it’s telling that it’s framed as a starting point rather than a mature strategy. Other government bodies, like the Government Digital Service (GDS), have been experimenting with AI for years, with projects like the GOV.UK chatbot showing tangible results. HMRC’s delay in reaching this stage suggests a lack of agility, perhaps rooted in bureaucratic inertia or resource constraints.

A Call for Bold Action

HMRC’s “landing zone” is a sign of progress, but it’s not enough. To truly harness AI’s potential, HMRC must move beyond pilot projects and incremental training. It needs a comprehensive AI strategy that prioritises rapid deployment, workforce upskilling, and collaboration with private-sector innovators. The government’s broader AI Opportunities Action Plan, which aims to make the UK an “AI superpower,” sets an ambitious tone. HMRC must align with this vision, not trail behind it.

The UK has a proud history of technological leadership, from Alan Turing’s foundational work to modern AI pioneers like DeepMind. HMRC, as a critical public institution, has a responsibility to embody this legacy. The establishment of a “landing zone” is a start, but it’s time for HMRC to stop taxiing on the runway and take flight. The longer it delays, the further it falls behind in a world where AI is no longer the future—it’s the present. 

Tax does have to be taxing.


HMRC Is Shite (www.hmrcisshite.com), also available via the domain www.hmrconline.com, is brought to you by www.kenfrost.com "The Living Brand"