TLDR: HM Revenue & Customs (HMRC) is under fire, accused of a ‘cover-up’ regarding the unauthorized use of artificial intelligence by its staff in assessing Research and Development (R&D) tax credit claims. This controversy follows a legal defeat in August 2025, which compelled the tax authority to disclose its AI practices, despite its denials of official generative AI use.
HM Revenue & Customs (HMRC) is currently embroiled in a significant controversy, facing accusations of a ‘cover-up’ concerning the unapproved deployment of artificial intelligence by its personnel. The allegations specifically point to the use of AI in the assessment and rejection of Research and Development (R&D) tax credit claims, sparking widespread concern among tax experts and businesses.
The dispute escalated following a pivotal legal defeat for HMRC in August 2025. A tax tribunal ordered the authority to reveal whether AI had been utilized in its R&D credit judgments. In its subsequent response, HMRC stated that its R&D compliance team did not use generative AI, further clarifying that such technology was ‘not approved for use in generating taxpayer letters’. The department reiterated that all enquiries are opened, managed, and decided by human officers, and any staff found misusing AI would face disciplinary action.
However, this carefully worded denial has drawn sharp criticism and skepticism. Advisers familiar with HMRC’s operations claim that, despite the lack of official policy, individual caseworkers did indeed employ AI tools to handle R&D claims as early as 2023. Sources, including one speaking to the Financial Times, indicated that ‘a number of people’ within HMRC’s small business compliance directorate were disciplined last year for using AI in correspondence. Another source suggested that while the practice has largely ceased, instances of ‘the odd caseworker’ using generative AI prompted HMRC leadership to implement new training on appropriate technology use.
Concerns are mounting over the potential ramifications of these unauthorized AI assessments. Businesses fear they may have been unfairly penalized based on AI-driven decisions, and there are significant worries about commercial confidentiality breaches, particularly if public AI models were used to process sensitive company data. Tom Elsbury, the tax expert who successfully brought the tribunal case, dismissed HMRC’s response as ‘smoke and mirrors,’ arguing it deliberately sidesteps the core issue of unapproved AI use. Richard Lewis of R&D claims company Pronovotech went further, labeling the situation a ‘potential cover-up,’ noting he had previously received a letter from HMRC in 2024 explicitly stating, ‘We do not use artificial intelligence to prepare our correspondence.’
The First-tier Tribunal’s ruling on August 2, 2025, which overturned earlier decisions by HMRC and the Information Commissioner’s Office (ICO), mandated HMRC to comply with Elsbury’s Freedom of Information (FOI) request by September 18, 2025. The tribunal found compelling arguments that HMRC’s failure to confirm or deny AI use reinforced suspicions, thereby undermining taxpayer trust and confidence and potentially discouraging legitimate R&D claims. This ruling highlights the critical importance of transparency in government use of AI, especially given the sensitive nature of tax data and the potential for national security risks if classified information related to defense or advanced technologies were mishandled by external AI platforms.
Also Read:
- Generative AI Adoption Trends Threaten to Exacerbate Gender Pay Gap in Key Industries, Research Warns
- The Perils of AI in Court: Why Self-Represented Litigants Risk Their Cases and Finances
This controversy unfolds amidst a broader crackdown by HMRC on fraud and error within R&D tax credits, a move that professional bodies contend has gone too far, inadvertently penalizing genuine claimants. The ongoing scrutiny underscores the challenges tax authorities face in integrating new technologies while maintaining public trust and ensuring accountability.
