State of use of AI tax systems

According to official and incidental mentions in literature of the OECD, Ireland’s Revenue Authorities have been using tax machine-learning algorithms since as early as 2011.

What functions are performed with AI?

Based on publicly available data, the tax machine-learning algorithms of the Irish Revenue Authorities perform three functions :

  1. Social Network Analysis (SNA): the SNA algorithm visually represents a network of individual taxpayers using graph theory, it represents a network of taxpayers as a combination of nodes for individuals or points of interests and lines which quantitatively and qualitatively measure relations between the nodes.

  2. Real-time risk detection (outlier detection): Outlier detection are unsupervised learning models designed to identify under-reported or non-reported income by comparing tax returns and tax documentation of an individual taxpayer with the tax returns of its peers, i.e. taxpayers in analogous financial and/or personal situations. The Irish outlier detection system uses multiple regression models based on Ireland’s income-consumption model which records patterns of consumption of the Irish population depending on certain levels of income.

  3. External risk-management (risk-scoring algorithm): The Irish Revenue Authorities use a machine-learning tool named REAP – risk, evaluation, analysis and profiling – which as the name indicates is an algorithm which evaluates and scores predicted risks of tax non-compliance or tax fraud and segments taxpayers into categories of risks, to subsequently select taxpayers for audits and develop an audit plan for the tax administration. A similar model is used for VAT purposes, see ‘VAT RTRF’.

What data can be processed by these systems?

The data processed for the development and use of these tax machine-learning algorithms is not specified. Yet according to statements of Michael Noonan, Ireland’s former Finance Minister, these tools can leverage all data available to Ireland’s Revenue Authorities. Section 872 (1-2) of the Taxes Consolidation Act of 1997 also provides that the data collected by the Revenue Authorities may be used for any purpose in connection with the assessment and collection of taxes, in any return, statements or documents of the Revenue Authorities.

Are these systems regulated by specific norms?

These tax machine-learning algorithms are not regulated by specific legal norms, and have so far been implemented solely on the basis of managerial decisions of the Revenue Authorities.


Legislation :

Section 872 (1-2) of the Taxes Consolidation Act of 1997.

Doctrine :