Overview

• A First-Tier Tribunal ruled HMRC must reveal details of its AI use by September 18, overturning the ICO’s earlier refusal to confirm or deny its applications in tax assessments.
• HMRC acknowledged it deploys AI to scan taxpayers’ social media for spending red flags exclusively in criminal investigations under legal oversight, insisting human judgment remains central.
• Internal planning documents outline expanded AI tools for routine operations, including chatbots to assist staff and taxpayers and automated “nudges” prompting suspected evaders to settle unpaid tax.
• Ministers and HMRC stress a human will always have final decision-making authority, but MPs warn that scanning personal online data risks errors and privacy breaches, citing the Post Office Horizon scandal.
• HMRC has approached around a dozen technology firms for AI solutions targeting the UK’s £47 billion tax gap, reflecting broader government plans to automate compliance checks and customer services.