Responsible AI in Government Tax Analytics and Compliance Systems

Artificial intelligence (AI), mixed with advanced big data analytics methods, has the potential to play a pivotal role in enhancing efficiency and accuracy in multiple tax processes, from data management and fraud detection to pricing optimisation and performance assessments.However, the use of these new technologies must respect the ethical principles of transparency, accountability, and explainability in order to secure their acceptance by both internal and external stakeholders, ensure compliance with increasingly stringent financial regulations and well-founded tax decisions, and ultimately facilitate a more equitable tax regime.A Responsible AI framework is articulated for an area typically overlooked in the discussion of the ethical use of AI in other fields: its application in government analytics and compliance systems for tax agencies, natural and legal persons subjected to taxation, and internationally coordinated data-sharing agreements.These institutions frequently combine their own datasets with information sourced externally through digital interceptions, companies and organisations mandated to carry out withholdings or disseminate internationally belonging to third parties, data of a fiscal nature of a different nature that guarantees compatibility in the absence of a tax treaty, and other types of financial data.Given the sensitive nature of tax-administrative data and the legal obligations in force for tax agencies, the AI systems developed should respond adequately to the three dimensions of Responsible AI: securing ownership of data and results; avoiding biases during the design stage and in the response stage; and guaranteeing control and security of the systems' outputs.