State of use of AI tax systems
The first mention of the development of tax machine-learning algorithms by the tax administration of the Republic of Slovakia data back to 2012, according to literature of IOTA.
What functions are performed with AI?
Based on publicly available data, tax machine-learning algorithms perform three functions for the Financial Directorate of the Slovak Republic:
- Taxpayer assistance – VCA chatbot ‘TAXANA’: TAXANA is a chatbot which can automatically answer taxpayer queries on various topics such as: tax returns and tax documentation, electronic filing, corporate, labour and personal tax burdens and VAT.
- Risk-detection ‘eKasa’: eKasa is a system of immobile and mobile electronic cash register, which transfers data to the tax administration in real-time, such as the unique identifier of the entrepreneur (UIK), the amount of the transaction, the time and place where the transaction was made, the price of the goods or services sold, the quantity of the goods or services sold, the price rounding, etc. (see below) The bulk of this data is then transferred to the central data warehouse of the tax administration for further processing by their risk-scoring algorithms. In real-time, the eKasa system can designate documents transferred as ‘invalid’, ‘deposit’ or ‘withdrawal’, which can be subsequently flagged as suspected transactions for further audit by tax officials by the ‘AIS-R’ system. As such, it is unclear whether the eKasa system is a simple financial ERP system (such as Oracle Financials), or has built-in machine-learning functions. Based on the legislation mentioned below, the former seems more likely. However, it is clear that the ‘eKasa’ ERP tool creates a system of machine-to-machine communication with machine-learning algorithms who further process said data, and thus it is important to mention such a system in the repository of aitaxadmin.eu. (For more information on eKasa see our dedicated publication on the case)
- External risk-management (risk-scoring) ‘AIS-R’: The tool predicts the risk of tax fraud or tax non-compliance associated with individual taxpayers to subsequently prioritise and (pre-)select taxpayers for further audits by tax officials.
What data can be processed by these systems?
The data used by TAXANA and the by the risk-scoring algorithms is not specified.
The data used in the eKasa system is listed in §8(1) of ‘THE LAW of 18 June 2008 on the use of the electronic cash register and on the amendment of the Act of the Slovak National Council no. 511/1992 Coll. on the administration of taxes and fees and on changes in the system of territorial financial authorities, as amended’, this includes: the electronic cash register code, the tax identification number, the unique personal identifier of the entrepreneur (UKI), the VAT identification number, the trademark, the seat of the business, the name of the business, the time and place where the goods or services were sold, the designation code of the goods or services, a description of the goods or services sold, the quantity of the goods or services sold, their price, the price rounding, the VAT rate and VAT base, exemption rates, and other documents for entrepreneurs in specific sectors listed in §8 (2) of the Law.
Are these systems regulated by specific norms?
Currently, TAXANA and the risk-scoring algorithms are not regulated by specific legal norms.
As mentioned the eKasa system is regulated by ‘The Law of 18 June 2008 on the use of the electronic cash register and on the amendment of the Act of the Slovak National Council no. 511/1992 Coll. on the administration of taxes and fees and on changes in the system of territorial financial authorities, as amended’
- The law of 18 June 2008 on the use of the electronic cash register and on the amendment of the Act of the Slovak National Council no. 511/1992 Coll. on the administration of taxes and fees and on changes in the system of territorial financial authorities;
- eKasa case: 492 FINDING Of the Constitutional Court of the Slovak Republic PL. ÚS 25 / 2019-117 (see our commentary on the case)
- IOTA, Data-Driven Tax Administrations (2012), pp, 14-15 & p. 41;
- M. Husovec (17 December 2021): https://husovec.eu/2021/12/the-slovak-constitutional-court-on-risk-profiling-and-automated-decision-making-by-the-tax-authority/