Landmark ruling on AI use in R&D tax credit decisions raises critical questions about transparency at HMRC and the effectiveness of AI assessors, warns Tom Haslehurst, director at Ryan
A groundbreaking tribunal ruling has forced HMRC to reveal whether it uses artificial intelligence (AI) in assessing research and development (R&D) tax relief claims. The First Tier Tribunal’s decision in Thomas Elsbury [2025] UKFTT 915 (GRC) has far reaching implications for businesses, taxpayers, and AI governance in public administration.
The tribunal case emerged from HMRC’s refusal to confirm or deny its use of AI systems in processing R&D tax credit applications. HMRC cited Freedom of Information Act exemptions designed to prevent tax fraud, but the tribunal found this position untenable, ruling that such opacity undermined public trust and could deter legitimate claimants.
The judge noted that concealing AI use risked undermining taxpayer trust and confidence in HMRC and could frustrate the scheme’s policy aims. The ruling highlighted suspicious indicators of automated decision-making, including correspondence containing American spellings.
